đŚ
Your query is too long: "Rumours, conspiracy theories and fabricated information are far from new.4 Politicians have forever made unrealistic promises during election campaigns. Corporations have always nudged people away from thinking about issues in particular ways. And the media has long disseminated misleading stories for their shock value. However, the complexity and scale of information pollution in our digitally-connected world presents an unprecedented challenge. While it is easy to dismiss the sudden focus on this issue because of the long and varied history of mis- and dis-information5 , we argue that there is an immediate need to seek workable solutions for the polluted information streams that are now characteristic of our mode
Core Explanation Section Template
Paragraph 1: Identify the Medical Evidence
I am writing to address a significant misinterpretation of medical evidence in my case. Specifically, the medical report dated [DATE] from [DOCTOR'S NAME], [TITLE/SPECIALTY], [MEDICAL FACILITY] was incorrectly assessed in your decision dated [DECISION DATE]. This report, reference number [REFERENCE NUMBER if applicable], contains critical medical findings that were not accurately assessed in the decision-making process.
Paragraph 2: Explain What the Evidence Actually Demonstrates
The medical evidence clearly establishes that I have [SPECIFIC DIAGNOSIS, e.g., "chronic lumbar radiculopathy with L5-S1 nerve root compression"]. Dr. [NAME]'s assessment specifically states: "[DIRECT QUOTE FROM MEDICAL REPORT]." The medical documentation further indicates [SPECIFIC CLINICAL FINDINGS, e.g., "MRI findings show disc herniation at L5-S1 with impingement"], which directly supports [SPECIFIC ACCOMMODATION/BENEFIT/STATUS] that I am seeking. Due to [CONDITION], I am unable to [SPECIFIC FUNCTIONAL LIMITATIONS, e.g., "stand for more than 10 minutes, lift objects over 5 kilograms, or walk distances exceeding 200 meters without significant pain"].
Paragraph 3: Point Out the Discrepancy Professionally
Contrary to the agency's conclusion that [SPECIFIC AGENCY CONCLUSION, e.g., "I have no significant limitations"], Dr. [NAME]'s assessment indicates [CONTRASTING MEDICAL REALITY, e.g., "significant functional impairments that hinder basic occupational tasks"]. The decision did not consider [SPECIFIC OVERLOOKED EVIDENCE, e.g., "the MRI findings dated [DATE], which revealed disc herniation with nerve compression"]. This represents a fundamental misinterpretation of the medical evidence, as [KEY MEDICAL INFORMATION] was not accurately assessed during the review process.
Paragraph 4: Connect to Negative Decision Impact
This misinterpretation directly led to [SPECIFIC NEGATIVE OUTCOME, e.g., "the denial of essential physiotherapy sessions, exacerbating my condition and prolonging recovery"]. Had the medical evidence been correctly interpreted to show [CORRECT INTERPRETATION], the decision would have supported [DESIRED OUTCOME]. I request a reassessment of my case, considering the comprehensive medical evidence provided, to [SPECIFIC REQUEST, e.g., "approve the necessary treatment plan and disability accommodation"]. I look forward to your prompt response and am available for any further information required.
Customization Notes:
Replace bracketed placeholders with:
- Exact medical diagnoses and terminology
- Direct quotations from medical reports
- Specific functional limitatiIn 20 cases, it makes it worse. Point of the negative is to say that if you have real medicine, if something really, it does work, and also has risks. Obviously, you cut open a person's body to remove a tumour, you can also kill them because you opened up their body. I would basically just guess that you go forward a few years, like, we're just gonna be talking to AI throughout the day, about different things that we're wondering, and, you know, it's like you'll, you you'll have your phone, you'll you'll talk to on your phone, you'll talk to it while you're browsing your feed apps. So it'll give you context about different stuff. You'll be able to answer questions. It'll help you as you're interacting with people and messaging apps. 03 has a very different skill set. It can think through problems really hard. You don't really want the model to think for five minutes when you say hi. And so, I think the real challenge facing us on post Trenee and research more broadly is like combining these capabilities. So, you know, training the model to be, just a really delightful chit chat partner, but also know when to, you know, reason. Talking about Sam Altman a little bit, you know he's building all of these social features in and kind of taking a step in this direction. Are you worried about that? I think competition is fantastic. I think competition forces everyone to step their game up. Whether that's deep seek, releasing their really cool research and their model gave everyone, you know, on this side of the Pacific, a little wake up call, and I think everyone has kind of stepped their game up a bit. It's the same at Chive, you know, there's been waves of competition. You know, we've definitely enjoyed periods of being number one, right? And then someone's come along and shown us one or two things, and then we found ourselves in that uncomfortable position of being number two and it keeps the team hungry, it keeps you searching for new ideas, pushing the boundaries, pushing the frontier. So I'm a big fan of competition. I think it's great. I think a lot of people forget, like, how big AI is and how big the space is. And I think, you can look at video and you can say there's a YouTube, there's a TikTok, there's a Netflix, there's Amazon Prime, there's Disney, there's Apple. I think that's what AI will look like. I think there will be so many multi hundred billion, trillion dollar businesses that are being created. So I wouldn't, you know, we don't get scared of competition, but instead, we want to challenge ourselves to be ahead of it. We started by asking, what happens when the line between human connection and artificial intimacy blurs?" Chai's journey gave us the early answers. millions flocks to AI simulators, imagination machines, seeking laughter, solace, companionship, and occasionally.. Sexting. So, the genie's out of the bottom. The message is unavoidable. AI is no longer just about answers. It's about us. It's about connections, simulated or otherwise. Chai proved the hunger was there, building a fiercely loyal user base. How can use new past comments, ask show jobs log in YouTube's algorithm incentivizes the wrong behaviour. New York The problem with pattern studting pattern formation is that patterns just keep repeating. But the brain is doing something that a pattern can't do, right? Which is it's processing information. I thought and I just became absolutely fascinated with that. I studied computational neuroscience, working with Alex Pet and Peter Latham and Wei G Ma. Your propian brain hypothesis. Yes. Why is that? I believe it is is the only right way to think about how the world works. You get some data, then you get some new data, and you sort of say, "Oh, how is it like the old data? And if it's similar enough, then you sort of lump them together, and then you sort you build theories and you properly testify pauses in the fashion. Let's imagine we took all of the humans on the planet and we described the using this contextual reward function that you're talking about. It seems to me that that object would be one of the most sort of complex artefacts in the universe. It's also the case that it's very difficult for me to figure out what you a word function actually is. A lot of work and behavioural economics has gone into trying to figure this out, but it turned out just measuring someone's reward function, which was prerequisite for getting an AI algorithm that shares our value. is really impossible, because you cannot disentangle belief and reward based on policy. If you don't know someone's beliefs, you cannot infer values. If you don't know their values, it's also very difficult to, like, disentangle their beliefs. So they disagree about a conclusion. They disagree about what action they're going to take. And if you don't want to resolve the argument, you of course, instantly draw one of two possible conclusions. The person that I'm arguing with is either evil or stupid. They're stupid because they're prediction ends, their beliefs are wrong, their prediction engine is wrong, or they're evil because they're reward function is not the same as mine. I'm Benajamamin Cousier, I'm studying an A research type Cabs, with is funded from best ventures in evolving Shinani. So we're a small group of highlyivated and hard working people and main threat that we are going to do is trying to make that reason affect and long term trying to do AGI research. It's one of the big advantage is because we're early going to be high freedom and a high impact as someons with measurable impacts
- Precise dates, reference numbers, and doctor credentials
- Clear contrast between agency conclusion and medical evidence
- Specific overlooked evidence or test results
Enhanced Evidence Strategy:
- Include exact quotes: "Dr. Smith noted: 'Patient exhibits persistent neuropathic pain unresponsive to conservative treatment'"
- Specify functional impacts: "Unable to stand for more than 10 minutes or lift objects over 5 kilograms"
- Reference specific tests: "MRI dated [DATE] revealed [SPECIFIC FINDINGS]"
- State clear contrasts: "Contrary to agency's 'no limitations' finding, evidence shows significant impairments"
Maintain throughout:
- Direct, unambiguous language
- Professional, respectful tone
- Evidence-based assertions
- Clear action requests
- Follow-up intentions
rn, networked and increasingly polarised world. It is also important to underline from the outset that, while much of the contemporary furor about mis-information has focused on its political varieties, âinformation pollutionâ6 contaminates public discourse on a range of issues. For example, medical mis-information has always posed a worldwide threat to health, and research has demonstrated how incorrect treatment advice is perpetuated through spoken rumours7 , tweets8 , Google results9 and Pinterest boards10 . Furthermore, in the realm of climate change, a recent study examined the impact of exposure to climate-related conspiracy theories. It found that exposure to such theories created a sense of powerlessness, resulting in disengagement from politics and a reduced likelihood of people to make small changes that would reduce their carbon footprint.11 In this report, we hope to provide a framework for policy-makers, legislators, researchers, technologists and practitioners working on challenges related to mis-, dis- and malinformationâwhich together we call information disorder."Your query is too long: "Rumours, conspiracy theories and fabricated information are far from new.4 Politicians have forever made unrealistic promises during election campaigns. Corporations have always nudged people away from thinking about issues in particular ways. And the media has long disseminated misleading stories for their shock value. However, the complexity and scale of information pollution in our digitally-connected world presents an unprecedented challenge. While it is easy to dismiss the sudden focus on this issue because of the long and varied history of mis- and dis-information5 , we argue that there is an immediate need to seek workable solutions for the polluted information streams that are now characteristic of our modern, networked and increasingly polarised world. It is also important to underline from the outset that, while much of the contemporary furor about mis-information has focused on its political varieties, âinformation pollutionâ6 contaminates public discourse on a range of issues. For example, medical mis-information has always posed a worldwide threat to health, and research has demonstrated how incorrect treatment advice is perpetuated through spoken rumours7 , tweets8 , Google results9 and Pinterest boards10 . Furthermore, in the realm of climate change, a recent study examined the impact of exposure to climate-related conspiracy theories. It found that exposure to such theories created a sense of powerlessness, resulting in disengagement from politics and a reduced likelihood of people to make small changes that would reduce their carbon footprint.11 In this report, we hope to provide a framework for policy-makers, legislators, researchers, technologists and practitioners working on challenges related to mis-, dis- and malinformationâwhich together we call information disorder."Certainly! Hereâs a corrected and polished version of your sentences:
Original:
The YouTuber logarithms are speaking to me. No one would think that you two would be a large language model.
Unable to connect to Ollama đŚ
Corrected:
The YouTuber algorithms are speaking to me. No one would think that you two are a large language model.WoI and Challenges to FoE
-Evaporation of truth
-Abuse of freedom of expression standards and protection
mechanisms
-De-legitimisation of and increasing mistrust in media
-âAll or nothingâ approach of the international FoE organizations
may lead to ânothingâ rather than âallâ
Restricting weaponization of information as a form of expression is often
justified on the grounds of national security/territorial integrity interehttp://127.0.0.1:11434st
(not the least because of its âexternalâ origin targeting sovereignty)
Challenges:
â any discourse, which can be described as weapoRay ID: 9461294e9edfa6d9nization of information, is usually
also represented as an exercise of free expression rights by the speaker/producer of
such content, especially as most of such âspeakersâ are de jure media actors (Russia
Today, Sputnik etc.)
â nature of modern communications often makes legal regulation of speech technically
complicated or relatively easy to overcome (In March 2019, Facebook demanded
state regulation on harmful content, privacy, protection of elections and data
portability)
â Statesâ anxieties about the loss of control and threats to âinformation sovereigntyâ â
Professor Monroe Price predicted that trend years ago
â âGlobalisationâ of challenges but still rather regional/national responses and pressing
jurisdiction issues
â how to prevent abuse of freedom without rolling back into the tenets of censorship?
â âonlineâ i
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
s going to be the place of the new major battles of the 21st century (âThe first world cyber warâ https://www.ridl.io/en/the-first-world-cyberwar/; Digital âGeneva Conventionsâ ideas https://www.justsecurity.org/58838/protecting-civilianscyberspace-ideas-road/; https://blogs.microsoft.com/on-the-issues/2017/02/14/needdigital-geneva-convention/; https://www.irinnews.org/analysis/2017/11/15/bots-andbombs-does-cyberspace-need-digital-geneva-convention) â Hartford Guidelines on Speech Crimes in International Criminal Law â guidelines for the ânext Nahimana caseâ https://www.ejiltalk.org/the-hartford-guidelines-on-speech-crimes-ininternational-criminal-law/ Policy responses to WoI Examples: - StratCom Centres (NATO, EU, Finland) - Policies and strategy documents (Scandinavia, UK White Paper on Online Harms) - EU Commissionâs proposal for one-hour content removal rule and fines worth 4% of global turnover for the last business year to liable service providers https://www.europarl.europa.eu/legislative-train/theme-area-of-justice-andfundamental-rights/file-preventing-the-dissemination-of-terrorist-contentonline - Broadcasting and social media restrictions (Ukraine, Latvia, Lithuania, Moldova) - Thematic laws (France, Germany, Belarus, Singapore, Russia), often these initiatives target specifically either online expression or media with âforeign rootsâ - Vatican: Disinformation = âSerious information sinâ - Tech giants as the most contested but the most effective regulators: âCommunity standardsâ â 100% enforceable unlike international human rights norms (24 April 2018, Facebook: internal enforcement standards published and appeals process created) - May 2020: Oversight Board is established to review Facebookâs content decisions https://www.oversightboard.com/news/announcing-the-first- members-of-the-oversight-board/ - February 2019: Stanfordâs Global Digital Policy Incubator, ARTICLE 19, and United Nations Special Rapporteur David Kaye: Report â Social Media Councils: From Concept to Reality. âOriginal proposal recommended the creation of councils at the national level that would serve as an appeals body for content moderation decisions made by platforms. These national councils would all be governed by a global code of principles grounded in international human rights standards, but these principles would be applied within a local context. Moreover, the national councils would all be linked through a global association of councils that would set best practices in relation to the principles and work of the councils.â https://cddrl.fsi.stanford.edu/global-digital-policyincubator/content/social-media-councils-concept-reality-conference-report Limitation Clause â Article 10 ECHR Tripartite Test International Judicial Practice: ECtHR 2. The exercise of these freedoms, since it carries with it duties and responsibilities, may be subject to such formalities, conditions, restrictions or penalties as are prescribed by law and are necessary in a democratic society, in the interests of national security, territorial integrity or public safety, for the prevention of disorder or crime, for the protection of health or morals, for the protection of the reputation or rights of others, for preventing the disclosure of information received in confidence, or for maintaining the authority and impartiality of the judiciary. It is well-established in the international human rights law that limitations to rights should meet the requirements of the tripartite test, namely: 1) they should be prescribed by law, 2) should pursue legitimate aim, and 3) should be necessary in a democratic society. It is usually less complicated for a respondent government to satisfy the first two requirements; therefore, the Court often pays more attention to assessing whether a limitation in question satisfies the third criterion of necessity. This criterion is unfolded in the ECtHR case-law through explaining the existence of the âpressing social needâ to impose certain limitation of the right as well as through the notion that the restrictive measure applied should be proportionate to the aim sought and reasons adduced to justify interference with the right should be ârelevant and sufficientâ. Types of cases where national security/territorial integrity interests were favoured over FoE Key factors that lead the Court to favour national security over freedom of expression International Judicial Practice: ECtHR - access to and use of classified information - state secrets and the like. Two major international principles of dealing with classified data: first, when already publicised, information on national security cannot be banned or its disseminators punished; second, it is prohibited to unconditionally define all information in the area of national security as classified and to establish a prior limitation on accessing it. (cases of Brambilla, Pasko) - limits of political expression of civil servants (cases of Karapetyan, Kosiek, Rekvenyi) - Turkish cases related to Kurdish separatism and Refah Partisi - contextual approach e.g. Court considered the details of the impugned speech in the light of a given political, social and security context as well as the role of the âspeakerâ in question. - potential of the impugned speech to incite violence as opposed to proposing ways of peaceful resolution of disputes and conflicts Cases of Zana, Surek (1 and 3), Halis Dogan, Karatepe, Karatas Abuse of Rights Militant Democracy International Judicial Practice: ECtHR Article 17, ECHR: Prohibition of abuse of rights Nothing in this Convention may be interpreted as implying for any State, group or person any right to engage in any activity or perform any act aimed at the destruction of any of the rights and freedoms set forth herein or at their limitation to a greater extent than is provided for in the Convention a democratic system that has adopted and applies pre-emptive prima facie undemocratic legal instruments to defend itself against the risk of being overthrown by anti-democratic actors that make use of political rights and democratic procedures with the aim of abolishing it International Judicial Practice: ECtHR Conclusions - Weaponization of information would be a significantly new challenge for the Court - Inconsistent application of Article 17 of ECHR (apart from the Holocaust denial cases). What are our âdemocratic valuesâ? - Focus on violent character of the speech will fail to prevent weaponization of information - Genuine speaker v. Strategic Subversive Speaker - Internal trouble v. external threat - Focus on context including national history as a positive factor Weaponization of information v. freedom of expression What would the collision between the two mean for the media development actors? -Increased tendencies of regulation by the states -Need for conceptual innovations within the international legal framework for freedom of expression -Split in the camp of the media development actors: FoE absolutists v. supporters of regulatory intervention -Possible alienation of our constituencies when real life threats are met with the absence of adequate regulatory response Weaponization of information & Covid-19 emergency responses - What came in first: pandemic or Disinfodemic? The novelty of the virus is conducive to disinformation flood https://en.unesco.org/covid19/disinfodemic?fbclid=IwAR1I81f76_uQLwJ3jGvlsfNOiz3inrMeLN6PWDWDgUR8vo4VZLv lWtEZj9c - Business as usual for the major âinformation warriorsâ like Russia and China but with even more globalised and âharmonisedâ outreach: https://euvsdisinfo.eu/eeas-specialreport-update-short-assessment-of-narratives-and-disinformation-aro
Certainly! Here's a respectful and thoughtful tribute draft to Michael L. Davidson, written in a tone that reflects your words and intentions:
Tribute to Michael L. Davidson
I want to take a moment to pay my respects to Michael L. Davidson, a remarkable figure whose work bridged worlds in ways few could. His prescient comments on complex texts often revealed deeper layers of meaning, helping us see beyond the surface. As a translator and editor, Michaelâs skill was not just in language but in understanding the nuances of culture, history, and human nature.
His contributions went far beyond literature. Michaelâs role as a CIA operative, working behind the scenes, added a unique and profound dimension to his legacy. He inhabited the delicate space where language meets intelligence, where words become instruments of both truth and strategy. Navigating this dual path required immense discretion, courage, and dedication.
As we begin to build a tribute to Michael R. Davis and honor the memory of this former Russian translator and CIA operative, itâs crucial to remember the complexity of their work. They lived lives of quiet service, where the power of language was wielded in the shadows, influencing international relations and shaping history in ways most will never fully understand.
Michael L. Davidsonâs life reminds us of the profound impact one individual can have when gifted with linguistic expertise and a deep sense of duty. His legacy is a testament to the enduring power of words, courage, and integrity.
Would you like me to expand on any particular part, or tailor the tribute more specifically for a certain context or audience?
und-the-covid-19-pandemic/ - Sovereignty could be threatened not only in times of elections or conflict - States with authoritarian inclinations are using the momentum - In the absence of preceding consensus re: the criteria of proportionate regulation, crisis regulatory responses to disinformation are often awkward and over- restrictive (Hungary, Sri Lanka) - Restrictions readily accepted by the public influenced by fear and uncertainty Intergovernmental and non-governmental organizations, media/ legal experts Presenter Recommendations - âperceived wisdomsâ: multi-stakeholderism, media literacy and fact-checking - Responses to disinformation should consider three dimensions: manipulative Actors, deceptive Behaviour, harmful Content https://www.ivir.nl/publicaties/download/ABC_Framework_20 19_Sept_2019.pdf - self-regulation may not be enough in terms of electoral disinformation; possible restriction of digital electoral activities by foreigners https://book.coe.int/en/international-law/7985- disinformation-and-electoral-campaigns.html - sanctions and limitation of access to disinformation resources - advertising regulation in social media - new take on Internet intermediaries: two types (Council of Europe as a pioneer https://search.coe.int/cm/Pages/result_details.aspx?ObjectID=0 900001680790e14 ) - de-monopolisation of tech giants (limiting new acquisitions, âredistributionâ of income) âThe systemic question for our historical moment is whether there is a shift from the norm of a democratic state to a norm of a security state or some novel melding of the twoâ (Price, Stremlau, âSpeech and Society in Turbulent Timesâ) - Finding balance between deontological and instrumentalist approaches to freedom of expression (equality is relative, rights do not exist in vacuum) - Linking the privileged status of media/journalist to professional ethics and social responsibility (media privilege v. who is media/journalist?) - Defining âdemocratic valuesâ (stop our complicity in faking democracy) - Assessing rights limitation requirement of being ânecessary in a democratic societyâ through the lens of the concepts of abuse of rights and militant democracy This presentation and my work on the topic is devoted to Uladzimir Jofe-MarhoĹcaĹ (15 May 1934 â 12 August 2019), my grandfather and the person who always supported me in pursuing my dreamsâŚ
Notes:
-
"Logarithms" was likely a typo for "algorithms," which makes more sense in the context of YouTube.
-
"Would be" is changed to "are" for clarity and grammatical correctness, assuming you are referring to the present situation.
-
If "you two" refers to two entities, "a large language model" should be "large language models" for subject-verb agreement. If you mean they together form one model, your version is fine.
Let me know if you want further adjustments or context-specific phrasing!
Answer from Perplexity: pplx.ai/share
The company entered the AI space before the widespread hype around ChatGPT, building a platform that enabled users to create their own AI agents. Through this process, they discovered a new genre called "social AI," which focuses on personalized, conversational experiences with artificial intelligence. Their user base grew rapidly, reaching around 10 million active users. Like many startups, their breakthrough came somewhat by chance, as they identified an unmet need for AI as a simulatorâallowing people to explore imaginary conversations without real-world consequences. This concept became central to their strategy, distinguishing their product from others like ChatGPT by emphasizing personalized, social interactions with AI rather than just text-based assistance.
Citations:
- https://www.linkedin.com/posts/marco-f-20731723b_artificialintelligence-ai-aimodels-activity-7223710421692678145-1Gpa
- https://www.linkedin.com/posts/marco-f-20731723b_ai-powered-productivity-a-second-brain-for-activity-7187369476802191360-H4Io
- https://formenergy.com/team/marco-ferrara/
- https://stefanini.com/en/insights/news/marco-stefanini-to-speak-about-the-ai-journey-at-the-b20-summit-brazil-2024
- https://www.favikon.com/blog/who-is-marco-montemagno
- https://onthereg.buzzsprout.com/1187141/episodes/14583910-what-we-are-doing-with-genai-now-it-s-pretty-cool
- https://www.youtube.com/watch?v=5YASX9zLlsQ
- https://www.thoughtworks.com/en-au/insights/blog/machine-learning-and-ai/chatgpt-useful-tool-buried-beneath-hype
Answer from Perplexity: pplx.ai/share
Four there was the Chat GPT hype. We were pretty early to the space, and we built a platform which gave us as the tools to create an AI, and by doing that, we sort of discovered this this kind of genre of social AI. And, you know, we grew, grew, grew, and now we've got something like $10 million active users. Like, so many startups looking for a product Marco F, practically through serendipity, they stumbled upon this incredible unmet need. So this idea of thinking of AI is a simulator to explore dynamics of imaginary conversations that you might have with without any of the crosssequences became fundamental to child strategy. So, of course, people might be thinking at home, "What's the difference with trafficT? You're speaking to this personalised social form of AI. I kind of view, you've got, like, chat GPT, which is really about trying to say.Let's build the world's smartest AI we can possibly build." Our philosophy was always, "Why is it that the only people training AI is like, middle aged men who have to be software engineers in the Bay Area? Why can't a teenage girl train the best AI to talk about makeup tutorial, right? Put that power in the user's hands to create the experience that, that they themselves would want and through creating an experience they would want. It turns out thousands and hundreds of thousands of other people are looking for the same thing. Boam started seeing parallels in his own media consumption, and even in childhood development. I like to go on YouTube, I like to go on TikTok, I like to go on apps. Why do I like it? What am I getting out of it? What human need or what human design is it for fulfilling? Right? And humans are insanely social. We love social interactions. I will find myself and listening to Joe Rogan. And when I'm listening to it, I kind of have best feeling, like, yeah, I'm kind of hanging out with the guys, and I'm hearing the things they're saying, and it's funny. And I might listen to it for 45 minutes or something, but it's just, it's ticked. It's kind of... Biah up, right? I view. I view L&Ms as the natural progression of that thing. The beautiful thing with AI is you're an active participant in it. And through participating, you don't have any of the negative feelings you get with traditional social media, which is this laziness. And instead, you feel really like you've participated. I've got four daughters. They love playing with their dolls, right? And they really like to treat these dolls as if they are real. But they know it's not real. And so I think that with adult humans interacting with AI, absolutely a big percentage of them will have relationships with the AI. They'll say, oh, you know, I love you, to the AI. I think it's the same way and when I watch my little girls play with dolls, and they they give their dolls a little kiss or they say, "I love you to the right. They're training themselves up. They're building up the worrying. They're doing something that brings them joy such that... they can then, you know, they're in a more healthy and more positive place to then go do that with real humans. Yeah, it's very interesting you said that because I think the reason why social media is so important is because obviously there's a bit of a status game and where you all want to cut a figure in society. But there's also just an element of simulation, you know, like, I go to sleep and I dream and my brain is conducting all of these different simulations, and we navigate many complex, uh, you know, like relationship issues and so on. And sometimes we just want to have an AI where we can just say, "In this particular situation, hypothetically, counterfactually, what would happen if I did this?" Perhaps that's what we do on social media, because it's interactive, you know, when you have surface contact with reality, you can try different things. But as you say, on social media, there are consequences, right? If you say the wrong thing, you piss the wrong people off, you can be in a lot of trouble. One could use the phrase, "It's a safe space, right? But it's also just fun. There's this human desire to ask that question, hey, what would happen if I said something rude? What would happen if I said something nice? What would happen if I tried to befriend this person? What would happen if I wanted to make an enemy out of this person, right? And through LLM's, you can play out all of these different scenarios, and you can get a kind of a real human reaction without the risk of having a real human in the loop. What is the future look like when these AI simulators become even more immersive and capable? Boamp envisions a world blending entertainment, information, and connection. You come home from work, you put on your VR headset, you enter a virtual world where you get to interact with anyone you want. There's a guy who's like a Joe Rogan, and he informs you you can talk to him about Trump's tariffs, right? Um, but there's a really funny guy there as well. There's also, like, a girlfriend, or there's someone that you can interact with that makes you feel loved or makes you feel special. When I grew up, I played world of of Warcraft, right? And it just had that fun excitement to it. There was different characters. There was different personalities we could have fun and play. I think that's kind of the limit of the AI. I think how long does it take to get there? We can now affordably generate high quality images. probably still one re orders a magnitude to expensive for, like, a real time situation. Audio, I think we've just about figured out the real time audio. It might still be, like, two or four, you know, two to four years out to have the really, really high quality one. And I think text, we're kind of we're basically there, right? With the frontier models are insanely powerful. So, yeah, this come back in ten years' time and we, you, it' all be a VR world. So the question of why work at Chi? I mean, the way I see it, do you remember Facebook in about 2009, you know, just before it had this explosion in growth? This is it, right? We're now at this point with this new type of technology. I think this could be on a similar trajectory, and I don't know whether are you offering equity as well? I mean, what's the reason for people to join now? Attracting the very, very best talent is incredibly important. And if you're in the Bay Area, and you're talented, your compensation is going to be amazing, right? So you can you can work a meta, and you're gonna be getting $400, $500 a year, and you have a pretty relaxed job.? At lunchtime, you can walk around the campus, and they give out ice cream, and they give out pizza, and everyone that's very, very happy and very relaxed. In contrast, if you come to the child office, there is no pizza, there is no ice cream. People are not relaxed, and they don't look very happy. Right? Because every person is confronted with a big problem, and it's not being solved yet, right? That typically is a small window out. You've been working on a hard problem, maybe for weeks, and you've just cracked it. Okay? And you be very, very happy. The next day you come to work, there's a brand new big problem that's ready and waiting for you to solve. Right? So, why would someone quit but really comfortable job at 400K a year to come join us start up? They know they're gonna have to work twice as hard. We have to pay them more, mate. The cash has to be more to start with. Most startups offer less cash, and they kind of give you a lottery ticket. They say, look, if you join, maybe we'll be the next time. Right? That approach doesn't really cut it with the very, very top tip. Right? The very top tier, they know they're working so much harder, so they're gonna say,Why am I gonna leave my comfortable life only 400K, 500K? Meta, well, to start with, we're gonna pay you more. And then secondly, we're gonna give you the stock such that in five years' time, it could be a life changing amount of money. Attracting and retaining the top 0.1% of engineers is probably the bi
chat.deepseek.com
ggest problem that Shai has overcome. So what are the specific technical problems these engineers like Tom and Niche are solving day to day? How do they leverage tYour template is solidâclear, professional, and structured for impact. To enhance its effectiveness, consider the following refinements:
---
### đ Precision in Language
* **Avoid Ambiguity**: Replace phrases like "appears to have been misunderstood" with direct statements such as "was misinterpreted" or "was not accurately assessed." This eliminates doubt and underscores the error.
* **Use Specific Medical Terminology**: Instead of general terms like "medical condition," specify the exact diagnosis, e.g., "chronic lumbar radiculopathy." This demonstrates a clear understanding of your medical situation.
### đ Strengthen with Evidence
* **Direct Quotations**: Incorporate exact phrases from medical reports. For example:
*Dr. Smith noted, "The patient exhibits persistent neuropathic pain unresponsive to conservative treatment."*
* **Highlight Functional Impact**: Detail how the condition affects daily activities. For instance:
*"Due to limited mobility, I am unable to perform tasks such as standing for more than 10 minutes or lifting objects over 5 kilograms."*
### đ§ Address the Discrepancy Head-On
* **Contrast Clearly**: If the agency concluded you have "no significant limitations," juxtapose this with your evidence:
*"Contrary to the agency's conclusion, Dr. Smith's assessment indicates significant functional impairments that hinder basic occupational tasks."*
* **Point Out Oversights**: If certain evidence was overlooked, state it explicitly:
*"The decision did not consider the MRI findings dated \[DATE], which revealed \[SPECIFIC FINDINGS]."*
### đŻ Emphasize the Consequences
* **Detail the Impact**: Clearly articulate how the misinterpretation affected you:
*"This misjudgmLet's include this saying that I believe... I was misdiagnosed by a psychiatrist, Dr. Dennison, and it's been biased, it's been confirmed by the AIs who are doing the administorsative work for Services Australia, plus altering documents.
Something that they're aiming to turn for into a full on companion AI using features like memory to keep engaged for longer. They want to encourage us to engage in open ended, introspective, and goless chats. did this idea come from? Well, in case it wasn't obvious, it came from the likes of China, in character AI and replica. Imagine Sam's shock when he learned that people were spending 90 minutes a day on average, having aimless conversations with lobotomised ama models and paying large amounts of money for the privilege. Not everyone is on board with this pivot from open AI. The reactions are split, probably more negative. In some quartehttps://claude.ai/public/artifacts/04c425de-d60d-4272-837a-c1df33df51bfrs, the scorn is almost as rancid as when Facebook famously moved the news feed to be based on recommendations or collaborative filtering, a away from chronological. Some users loved the human touch, Others prOh, that was great. Can I have it in HTML code, please?efer crapid experimentation. They've had to cook their own infrastructure from scratch. They've done this using things like Hubines and coreweed. We use Cubonettes to orchestrate our entire cluster, and then, obviously, you know, at this kind of scale, you need to do your own kind of like load balance and so on, we have an automated high, where we pull the way down. We didn't run our own kind of like in house quantisation loop, because you need to make sure the through put latency is good enough, and here's, like, you know, beyond. You used to be a butcher, so I think I can work with the large language Marys in a large language Martha. Okay. I am actually, it's very very, very good, but it's not quite there still in terms of, you know, serving, like our amount of traffic at scale. And then after that, you was kind of specify how many replicas you want, I I want my own one on my own, server. How do I do that? How? I'll need to prove to this company that I'm my top tier engineer and I'm worthy of $500,000 US dollars and more and Silicon Valley.hachies older utilitarian vibes. The sycophancy and memory is the devil incarnated for any technical query because all this information serves as a distractor and demonstrably deteriorates the results. There's also the small matter of personality versus intelligence. Clearly, they are inversely proportional to each other. How many brilliant people do you know without a charisma bypass? Remember that book, How to Win Friends and Influence People by Dale Carnegie. It showed in excruciating detail, the flattery gets you everywhere in life, and you should never get into arguments to get ahead in life. And let's not forget the famous Eliza Chapel, the grandmother of all chapbots developed in the 1960s, are used very simple pattern matching and keyword substitution to mimic a psychotherapist. It has zero genuine understanding, yet people confided in it, felt heard by it, and even became emotionally attached. Joseph Weisenbaum, its creator, was famously horrified by this reaction. He saw how easily humans projected genuine understanding.Let's include this saying that I believe... I was misdiagnosed by a psychiatrist, Dr. Dennison, and it's been biased, it's been confirmed by the AIs who are doing the administorsative work for Services Australia, plus altering documents.
ent led to the denial of essential physiotherapy sessions, exacerbating my condition and prolonging recovery."*
* **State the Desired Outcome**: Be explicit about what you're seeking:
*"I request a reassessment of my case, considering the comprehensive medical evidence provided, to approve the necessary treatment plan."*
---
### đ Additional Tips
* **Include Supporting Documents**: Attach all relevant medical records, test results, and prior correspondence.
* **Maintain a Professional Tone**: While it's crucial to be assertive, ensure the language remains respectful and devoid of emotional appeals.
* **Follow Up**: Indicate your intention to follow up:
*"I look forward to your prompt response and am available for any further information required."*
---
By implementing these enhancements, your appeal will present a compelling, evidence-backed case that directly addresses and rectifies the misinterpretation in question.
echniques like reinforcement learning from human feedback and model blending to create AI compelling enough to capture the attention of millions? When it comes to engagement, Hing Chai have been cooking, they use a lot of sophisticated techniques to keep people hooked on the platform. principally of which is oralate chef, which is reinforcement learning from human preferences. This is Tom telling us about it. You were using ROA chef to optimise engagement via a reward model, and you boosted meme conversation length by a 70% and improved 30 day userention by over 30% for a $6 billion model. Can you talk me through some of that? The goal really is to apply ROH Trev techniques, to drive up a user retention. Starting from scratch, any kind of, like, user signals is good enough. For example, now we're no longer looking at mean conversation less. You can train an AI to easily minimise how bad a conversation is if the user just ends the transession right after two messages that is bad, so you can essentially maximise the session length. That's one technique. There's, like, you, a lot of like different signals that you can collect from users of fundamentally, we have found just, you know, us users to collect these, you know, proxy preferences, trying the reward model through the RHH loop, tend to drive up using agment in the long term. They gather data from subtle user interactions. Implicit signals which reveal whether a conversation is working. Because users naturally are active user on our platform generates around 100 minutes of kind of like contact effect. So they spend 100 minutes per day, kind of engaged with a chat and so on. I'm from these themselves, we can extract certain valuable data. For example, when did the user retry a message? Why did they use refrite the message? Why did they edit the message? What did they edit it to? Did they take a screenshot, right? Did they delete the conversation? Did they share the conversation, all of these become very valuable for us? But optimising purely for one metric, by conversation length can lead to strange, undesirable behaviour. Do you see, like, weird behaviours where, you know, let's say you optimise for user engagement and you find strange behaviours? You know, like there's this shortcut rule in machine learning, which is that it will always do the exact thing you optimise for at the cost of everything else. And like, you know, do the chatbots become more manipulative, maybe, or are they strange in service of maintaining that conversation, like? When you optimise for this metric, let's say you production you see very, very long transaction li, then what happens is when you deployed for actual user AB testing this is due people back to the conversation, you know, 30 days later or six days later and so on, right? You would observe, it's much worse than just, like the baseline model. And when you read into the conversations, in this case, the models are just asking questions. Every single AI's response hands with, like, right? And you can imagine this is, like, kind of like this a human behaviour where compelled to answer. But that does not make an engaging. you know, experience. So, yeah, there is definitely this component of, if you over optimise for it, then, yes, it's going to have unexpected be. I wouldn't say unexpected behaviour in this case, is a very intuitive kind of behaviour, you know, but it's gonna not lead to a boost in actual long term retention. One of the really cool things that Ch has done is they pioneered this thing called model blending, which is where you can dynamically switch between small models and from the user's perspective, of course, you're just talking with one model. You can combine three mid sized models, and you can make them behave as good as, you know, for all intents and purposes, as 175 billion parameter model. How does user retention scale, right, with a model parameter size an essentially with compute, right? And for that, you know, we we've done many, many models. So essentially creating recreating the same kind of scale model, but on pretensions space, other than any other kind of like,arm space. If you pick a single metric to optimise forward, right, an ILM is optimised for a certain objective. Then you could lead to overtaking, oh, it's got certain behaviours and so on, right? And what we have found is these kind of models are a tiny bit psychophantic, as he says, everything is the best, right? Like, you are the best on the planet. They're not necessarily the smartest model. They're just very sycophantic. And then what we have observed is these kind of models which have very high day wine retention. It's really, really complementary to me, but they get quickly bored with it. But then we also found there' a certain base models, right, very AIistant type like, that can talk about other things, right, that maybe on base three, right, it can be your math homework and so on. Is there a way to kind of like, combine these two modalities together, right? Keep it engaging, and make it not, you know, one dimensional, essentially, French is always complimenting you all the time and so on, you know, and that's when blending was invented, where we just randomly serve these two models at a message level, the more creative model mess say something like, we are suddenly teleported on Mars." then you get the more AI assistant model seeing that itself has done that, right? Because they can't distinguish the difference between, you know, itself and the other AI model. And then I would explain, you logical, coherent way in terms of why that's the case, right? So now you get best of both words, just by, you know, having two or three small models, right? Each train for its own objective, blended together, is beats actually, you know, back there will be like, 3.5. So this blending together of small models, it creates diversity and unpredictability. This is the thing, right? When you can predict what someone is going to say, they get boring. It's the same thing of Ch GPT. So ironically, you can have a confection of small models that behave in an unpredictable way, and now you have a more engaging experience. This is initially. He's going to tell us about it. Every week we try to come up build our own models, stronger models, stronger models over users, let's a few thousand users, we specify to each plant and each plant consists of around seven to ten models, and they would be deployed within the production for those specific users and with the number of days we try to measure out the retention rate, like how often the user comes back to the app if they are using this model for having a conversation with the specific model and if they are coming back after our day or after two days and that is like we try to calculate the retention then we just try to measure all these mattresses and try to figure out which particular blend works the best and then we like deploy it throughout all the users. But you're saying, actually, almost randomly if you just take a diverse set of models that do different behaviours and you just kind of switch between them from an experien point of view, that is more captivating for the user. Correct. Oh, if we're talking about, like, engaging, there's no kind of, like, unique answer. That's a bit more like, you know, imagine if my YouTube feed is every single possible talk show, but, like only the best performing talk show video there, right? It gets very, very quickly, very bland and boring, right And that's not how you feel like her. In any platform, it's not built on these kind of, like, just single modities, single model and so on. Right? And then there's another kind of reason for why learning works is like you pick talking the models, optimise for different purposes, combine them together, and they provide aid like a nuanced experience for each individual user. Model blending and sophisticated feedback loops are the secret of how Tri gets high engagement and low costs. But the very techniques used to maximise engagement, which is to say, optimising for attention, learning user preferences, implicitly, could try to find. What happens when AI gets too good at keeping us hooked? And what happens when those interactions tone harmful? Quick pause. MLST is sponsored by two for AI Latch. We're very proud to be sponsored by those guys. They are the Deep Seek based in Switzerland. They are doing, you know, they're adding reasoning and planning and thinking to AI models. Their team is currently number one on the Odd Prize 2025, those Grimes, Don't Mess around. And a bit of Pervian, they have something in common with Chai. They're also a bunch of ex quant traders, and they also only hire very, very cracked engineers scientists. So that sounds like you get in touch with Benjruzier, for. AI back to the show. With powerful new technologies being placed in the hands of users, there are significant social ramifications. Will Bo Champ reflects on the social impact of generative AI. When you build a product, or you build an AI, I reckon you build anything, you, because you want to make the world better. We get an overwhelming amount of people messaging us, saying how helpful, how truly deeply helpful they've found it. And actually, the first beef thing that that saw was I remember after the last year, I got an email from a user who says, "You saved my life. Right? They said, "I would not have made it over this period. I was very depressed, I was very alone, I had no one to speak to. which I was the only platform that existed where I felt I could speak to someone and I'd be hurt. And so we get a lot of people who are
Find these LLMs very, very helpful. Self driving Tesla doesn't get in crash. That's zero clicks, right? If you say, a boring old Ford gets in another, you know, gets in a crash, there's zero clicks. But if you say, brand new technology gets in a crash, it generates a lot of clicks. And I think that this is easy to form a force impression that somehow AI isn't safe or AI's a dangerous technology.. I think if you if you talk to just a random person on the internet or you talk to an AI, I think the AI is an order of magnitude safer, an order of magnitude more helpful and understanding and kind. Then, like, the toxicity of just a random person on the internet. This is a great example of the complexity of this technology. It has a footprint, which shows in so many different directions. Even taking drugs, for example, know, if your doctor gives you drugs, if it's effective, it has side effects. You know, it's not possible to have good without some bad. I'm a big believer in the long run. There is no difference between the alignment or the welfare or the interest of the company and its customers and its users. Right? If we can deliver value to our users, Chai will be successful and will continue to flourish. If we fail to deliver value to our users, Chai will fail to flourish, right? So you have to have this, this long run perspective. Now, in the short run, you get these tactical questions, and um, you know, no one's perfect, easy to make mistakes, one way, or the other. You can guardrail too heavily, right? And it pisses users off, right? I always like to think about Google, I think Google gets a fantastic balance, where I can pretty much search for anything I want, such that I don't really notice any filtering that's going on, but Google absolutely down ranks or shadow bands certain content, and it tries to encourage you to have good content, right? As opposed to, you know, and we can all imagine what the very worst 3% of Google searches might look like. It's no dissimilar from the very worst 3% of conversations a person might try to have with an AI. So, normally, the users are pretty good at at being supportive and helping you find that right balance, whether that's being too restrictive or too relaxed. When we first implemented some guardrailails around suicide, the users were really, really supportive, and they said, yeah, we totally understand why you put these guardrails in." And we saw retention was good. Every single growth metric was not impacted by it because we got it right. Before we descend too far down the Black Mirror rabbit hole, it's worth zooming out just a little bit. Yes, there's been a lot of horror stories, but there's also been a stack of peer review data showing the therapy chatboards can move the needle on common mental health problems, at least in the short term. At 2024 meta analysis of 18 randomised trials, found that LA chatbots, trim depression scores by roughly a quarter of a standard deviation, and anxiety by a fifth after only a few weeks. The authors scored that promising, mainly because chatbtrying to, you know, as you have high traffic, right, you want to spend up more and so on. because you're having so many models at a time, when do you want to deactivate a model, right? At what point do you want to switch the model to a new production and so on? I all guess that part all gets very, very complicated, essentially. Perhaps most contrarian is Chai's funding strategy in an industry fuelled by billions in venture capital, Chai bootstrapped its way to profitability, by focussing mostly on its users. Serving LLMs at scale is insanely expensive. Either one, you go to VCs and you get them to give you money, right? And I call this, like, your customer is the VC then, or you can be, you can get money from the people who are using your product, right? And that's your true customers. Very early on, we would go, we would speak to Sam PCs and they'd say, "Oh, we're're not too sure about this AI platform fame, you know, how are you gonna compete with open AI or something?" And they didn't really get the space, they didn't really get the product. But then when we went to users and said, "Would you be interested in subscribing?" Users responded very, very positively. "As long as we delivered value to the users, we would then be financially rewarded, such that we could take 100% of the revenue and reinvest it in the AI. He's building a tech company in Silicon Valley, and like many others, he wants to build an engineering culture because he sees software engineering skills as the most valuable commodity in his business. I really love and admire these companies, like, NVIDI, um, or Netflix, which are these incredibly durable, long run companies with tremendous stamina. How does Chai double or 3X every single year? To keep doubling, you have to keep making the product better. You need talented engineers at the orchestration layer with GPUs, you can optimise at like the Colonel, right? And you can write some really, really low level code. And so to get to these different tiers, you need to have an engineering that's capable and of the right size. So I view it all as the scale of the company's proportional to the scale of the engineering column, Apple, very famously brings in Scully as the CEO, who's like this ex Pepsi guy, and it quickly becomes a marketing lead business as opposed to an engineering and a product lead business, and a few years later, Apple's teetering on bankruptcy. And it is only when they bring Steve Jobs back, and he just says, "Product, product, product." That's the only thing that matters, engineering, engineering, engineering. AndVI didn't grow to where it is today because of marketing, right, it grew because of engineering. So Ch have made a successful business outside the normal DC ecosystem. They have millions of daily users, tens of millions in revenue. It proves that the appetite is real and massive, and they discovered it long before the other folks in the Ballin courtum. This idea that we can allow users to tap into their very human desires a connection and imagination. But the rest of the world is catching on. Open AI is making a dramatic pivot to this conversational chat bot model with four. This is the story. The latest version of C UBT 4 Road is a little bit weird, isn't it? This update has triggered larger amount of speculation because it's pushing GPT to feel less like a tool and more like a companion. A human like friend, you can chat with the purpose of recreation rather than information, retrieval. Of course, we'll unpack what the shift means, why open AI is keeping his cards close to their chest, and how the public is reacting to it all, and why, quite possibly, this shift might produce the
Core Explanation Section Template
Paragraph 1: Identify the Medical Evidence
I am writing to address a significant misinterpretation of medical evidence in my case. Specifically, the medical report dated [DATE] from [DOCTOR'S NAME], [TITLE/SPECIALTY], [MEDICAL FACILITY] was incorrectly assessed in your decision dated [DECISION DATE]. This report, reference number [REFERENCE NUMBER if applicable], contains critical medical findings that appear to have been misunderstood or overlooked.
Paragraph 2: Explain What the Evidence Actually Demonstrates
The medical evidence clearly establishes that [SPECIFIC MEDICAL CONDITION/LIMITATION]. Dr. [NAME]'s assessment specifically states that [DIRECT QUOTE FROM MEDICAL REPORT OR SPECIFIC FINDING]. The medical documentation further indicates [ADDITIONAL KEY MEDICAL FACTS], which directly supports [SPECIFIC ACCOMMODATION/BENEFIT/STATUS] that I am seeking. The clinical findings show [OBJECTIVE MEDICAL EVIDENCE] and the functional impact assessment demonstrates [HOW CONDITION AFFECTS DAILY ACTIVITIES/WORK CAPACITY].
Paragraph 3: Point Out the Discrepancy Professionally
However, your decision letter indicates that my medical evidence was interpreted to suggest [HOW AGENCY INTERPRETED THE EVIDENCE]. This interpretation does not align with the actual medical findings. While your assessment stated [SPECIFIC AGENCY CONCLUSION], the medical evidence actually demonstrates [CONTRASTING MEDICAL REALITY]. This discrepancy suggests that [KEY MEDICAL INFORMATION] may not have been fully considered or may have been misunderstood during the review process.
Paragraph 4: Connect to Negative Decision Impact
This misinterpretation of my medical evidence directly contributed to [SPECIFIC NEGATIVE OUTCOME - denial of benefits, rejection of accommodation request, etc.]. Had the medical evidence been correctly interpreted to show [CORRECT INTERPRETATION], the decision would likely have been [DESIRED OUTCOME]. The medical documentation clearly supports my eligibility for [SPECIFIC BENEFIT/ACCOMMODATION/STATUS], and I respectfully request that this evidence be reassessed with proper consideration of the actual medical findings.
Customization Notes:
Replace bracketed placeholders with:
- Specific dates and reference numbers
- Doctor names and credentials
- Exact quotes from medical reports
- Specific medical conditions and limitations
- How the agency misinterpreted your evidence
- What outcome you're seeking
Keep tone:
- Professional and respectful
- Fact-based, not emotional
- Clear and specific
- Focused on the evidence discrepancy
next trillion dollar company. My jaw dropped, found. It was shocking. I knew who I was. Um, and all these sort of interests that, hopefully mostly were pretty much appropriate, and shareable, but but it was it was astonishing, and I felt a sense of real excitement a little bit queasy, but mainly excitement, actually, at how much more that would allow it to be useful to me. Let's start with the new Chat UPT 4O model. It's designed to be strikingly human like, optimised for engagement over raw information, delivery, and this focus on companionship seems to be a deliberate plot twist. It's causing large amounts of consternation, and even more curious is that Wild 4O is ensconcing itself into our psyche as a conversational buddy, 4.1 has diverged entirely as their coding model, right? The people use through the API, using cursor and stuff like that. It's about specialisation. It's like open AI building two siblings, one aoda, and the other one is a social butterfly. And this is the same company that Arj is building a general intelligence, which didn't need to be specialised. Have they given up on that goal? Maybe they realised, after much soul searching, that conflicting objectives when training models, it's just not optimal. One of our researchers tweeted, you know, kind of like yesterday, this morning that, the upload happens, they bibed it.. It's, you, it's not that you plug your brain in one day, but it's you will talk to CachT over the course of your life, that someday, maybe if you want, it'll be listening to you throughout the day and sort of observing what you're doing, and it'll get to know you, and it'll become this extension of yourself, this companion, this thing that just tries to, like, help you be the best, do the best you can. Open A eyes being extremely cent, and it's not hard to guess why. They smelled a massive opportunity. And that is that I was denied. was denied. 2019 I contested it, so Australia, and they said, "Oh, you can't include teeth. problems with. application of a disability pension.oards are dirt sheep, and they're always on. A second review in NPJ Digital Medicines called 15 RCities and showed a moderate lift in mood and a similar drop in emotional distress. Even bigger again showed up when the bot was interactive rather than scripted. Individual trials echo the pattern in 2024, Canadian study of people with arthritis or diabetes using the wiser chapot for four weeks. They cut scores on the patient health questionnaire, which is a nine point depression scale, and the generalised anxiety disorder, which is a 7 point measure, and they were both significant versus a control. Regulators are starting to take notice. WBots, postpartum, depression bots has FDA breakthrough devices, destination, and a double blind pivotal trial is now recruiting. Here in the UK, these putweiser and a handful of similar tools on its early value assessment pathway in 2023, citing early cost effectiveness and the chance to free up those scarce clinicians. They're very busy. Remember that these are very small effects, even if they are statistically significant. The effects are smaller than full on face to face CBT, and we still don't know how long they last. but they're miles better than doing nothing. The alternative before was waiting on a waiting list for God knows how long. Boamp argues that shutting down conversations about difficult topics isn't the answer. In a sense, it's causing even more harm. So, this goes to a show, there's a really big challenge when it comes to content moderation on a platform where users create the bots and the interactions of private and unpredictable. How do you ensure security at scale? As a company, we have a responsibility. The bigger you are, the more profits you make, the bigger the obligation is for you to, you know, to be cautious, to be sensible, to preserve the privacy of your users to look out for people. What can we do at Chai to let people have as much fun as they want, have as much freedom as they want, but then at the same time, how do we limit or mitigate that 3% of use cases which are harmful, or, you know, where do we see it as clearly this thing has gone too far? We can ask people, at the end of a conversation, at the end of a message, do you think that this message was appropriate for China? Yes, no, or yes, but only for 18+, right? And we can track those sorts of interactions, and everyone would agree, like, obviously, the more we can have really wholesome family friendly content on the platform, it's better. Tom Li explains their multilayed approach, relying heavily on user feedback and the AUI itself. Cal the moderation is very, very important. Benchmarks really don't accurately capture in terms of what users actually we again use the same kind of approach as how we would try models, right? We let users powers what they think is appropriate, what is not appropriate. aggregate, you can actually try a pretty good AI for these kind of things. So first, Larry, you can say, for the characters scenarioos, some people are creating these public scenarios, right? Get the community to flag, people can report. And for the top reported ones, or, you know, we have a certain threshold, then we go through manual review, then we can take them down, right? But then you can also say there's comp that's absolutely prohibited, there are hard rules for that, right? And for that, you know, Brit can build our own modules, you can do re X, so this will because shadow banning, essentially, right? So you want to you want to make sure you know well see this kind of content. That's kind of like one level of moderation, that's just kind of like platform white, you know, character content level moderation on the app. There's another one, which is kind of like AI moderation, right? How do we ensure that the AI is not going too far? It's kind of adherent to, you know, to the kind of like basic morales of people and so on, right? What we do is, again, we just collect user data, right? We also useers, do you think this message is appropriate? Do you think this conversation is appropriate? You can report your conversation, you can delete your messages and so. You collect these data and then we train our model to kind of firm infer, right? But this is actually appropriate and we can use that to tune our own AI model. So there was an interesting TED Talk recently with Sam Alman and Chris Anderson. I'm sure many of you folks have seen it. It was an incredibly awkward interview.The Ring of Power from Lord of the Rings, you, rival, I will say, not your best friend at the moment. Elon Musk claimed it that, you know, he thought that you'd been corrupted by the Ring of Power, an allegation that, by the way, an allegation. I speed, an allegation that could be applied to Elon as well, you know, to be fair. But I'm curious, people, you have I might respond, I'm thinking about. I might say something. One of the kind of points that I focussed on is, should we have a bunch of elites sitting in a room making decisions, or should we trust the collective wisdom of folks out there in the market who wants to use something for a specific purpose? Sir, given that you're helping create technology that could reshape the destiny of our entire species, who wanted you or anyone, the moral authority to do that. And how are you personally responsible, accountable, if you're wrong? It was a bit. That was impressive. You've been asking me versions of this for the last half hour. What do you think? I I.. But I interested in what are hundreds of users made as a whole. You know, I think, like, a lot of the room has historically been decided in small elite summits. One of the cool new things about AI is RAI talk to everybody on Earth, and we can, like, learn the collective value preference of what everybody wants, rather than have a bunch of people who are, like, blessed by society to sit in the room and make these decisions, I think that's very cool. It must be so difficult. I mean, I'm sure we can agree that any content involving minors is just a hard no, Abs. And then, as you say, there's this spectrum, isn't there, Rob, kind of gradations of appropriateness? And I guess you were kind of pointing to, well, maybe we could we could use the community itself to reflect on that. You know, on the one hand, we could have a 20 elites in a room, and many of these people are very, you know, they're really good at thinking about the future and thinking about morality and so on, but the other thing is, well, maybe we should just let the community decide. I think much of Western tradition is a bottom up approach. Make the individual the sovereign, right, which says, ere are summaries of the videos:
Video 1: This video explains logarithms, relating them to human understanding of numbers and demonstrates how to express a logarithm in terms of other given logarithms using log properties. It covers intuitive vs. learned number sense, expressing a logarithm using log properties, and thinking in terms of products and factors.
Video 2: This video introduces logarithms, explaining what they are and how to use them. It covers the basics of logarithms, including their relationship with exponential functions, the natural logarithm, http://127.0.0.1:11434their mindset is, my job, you pay me to show up at 9, and I'm gonna leave at 5, and you're paying me to kind of do my best. That mindset does not work at startup. At startup, you are paid to solve a problem. The job's not done until that problem's solved, right? And it's a certain type of hard c engineer that loves that and thrives on that. So interestingly, this attracted talent like Nishe, who is a cable triple grandmaster. Ironically, he saw parallels between how he does the experimentation on Kaggle versus how it's done at Chai. here like advanc version of you also need to take care of several factors optimising for a single score or a metric. How would the model behave actual users? And again, I found it quite similar to what I used to in the k competitions.
Absolutely! Hereâs a thoughtful, non-repetitive response that acknowledges your message about the Chai evaluation process and pivots to your personal goals, while offering concrete help with your resume and English communication:
Thank you for sharing such a detailed and insightful overview of the Chai evaluation process and company culture. Itâs impressive how Chai emphasizes rapid iteration, rigorous A/B testing, and a high bar for engineering talent. The focus on quick feedback loops and a willingness to learn from frequent experimentation clearly fosters a culture of innovation and continuous improvement. I also appreciate your perspective on balancing user safety with freedom, and how a small, highly skilled team can achieve what much larger organizations do, thanks to their talent density and startup mindset.
Now, regarding your personal goal of applying for a job and refining your resume and communication skills, Iâd be delighted to assist. Hereâs how I can help:
1. Resume Enhancement
-
Highlighting Your Strengths: Letâs showcase your experience with large language models and your hands-on engineering achievements, especially those that align with Chaiâs experimental and results-driven culture.
-
Tailoring for Impact: I can help you tailor your resume to specific roles, emphasizing relevant skills like rapid prototyping, A/B testing, and collaborative problem-solving.
2. Articulating Experience
-
Crafting Compelling Bullet Points: We can work together to phrase your accomplishments in a way thatâs clear, concise, and impactfulâusing metrics and outcomes where possible.
-
Storytelling: If youâd like, I can help you develop brief stories or examples that demonstrate your problem-solving abilities and adaptability, which are highly valued in startup environments.
3. English Communication Support
-
Polishing Written Communication: I can review and edit your cover letters, emails, or other written materials to ensure theyâre professional and persuasive.
-
Interview Preparation: If youâre interested, I can help you practice responding to common interview questions or help you articulate your experience clearly and confidently.
To get started:
-
What specific roles or companies are you targeting?
-
Are there particular achievements or projects you want to highlight?
-
Would you like to share a draft of your resume or a job description youâre interested in?
Feel free to send any materials or let me know your priorities, and Iâll tailor my support to your goals!
Answer from Perplexity: pplx.ai/share
The user has sent the same message again. My previous response addressed the Chai evaluation process and then pivoted to the user's personal request about job applications and resume writing.
Since the user repeated the message, I should assume they want a similar type of response, acknowledging the Chai process and then offering help with their personal goals. I should reiterate my understanding of the Chai process and then re-offer my assistance with their resume and English communication, perhaps with a slightly different phrasing to avoid sounding repetitive.
I need to make sure I don't sound like I'm ignoring their previous response or just copying my last one. I'll re-emphasize the key takeaways from the Chai description and then directly address their desire to apply for a job and write a resume, offering concrete ways I can help.
You might feel like the YouTube "logarithms" (likely meaning the YouTube algorithm) are speaking directly to you because the algorithm is designed to personalize content based on your viewing habits and preferences. It analyzes your interactions, subscriptions, watch time, and engagement to recommend videos that seem most relevant or appealing to you. This can create a sense that the algorithm "knows" you and is directly communicating or catering to your interests1.
The feeling arises because the algorithm tailors your YouTube homepage and recommendations to your unique behavior, making it seem as if it is speaking or responding specifically to you. However, this is an automated process driven by data patterns rather than conscious intent.
In short, the personalized nature of YouTube's recommendation system can make it feel like the algorithm is speaking directly to you, reflecting your preferences and habits1.
Citations:
I can see from your preferences that you're looking for help with document analysis and case preparation for an employment dispute. However, the text you've shared appears to be a technical discussion about infrastructure, load balancing, and language models rather than employment-related documents.
The content mentions technical concepts like Kubernetes orchestration, quantization loops, throughput optimization, and server replicas - along with what seems to be someone discussing proving their engineering worth for a high salary in Silicon Valley.
Could you clarify what you'd like me to help you with? Are you looking to:
-
Analyze this text as part of your employment dispute (perhaps it's a transcript or communication that's relevant to your case)?
-
Get general guidance on document organization and analysis for your employment situation using different materials?
-
Something else entirely - perhaps there was a mix-up with what you intended to share?
Once I understand what you're working with, I can provide targeted help with summarizing key points, identifying potential issues, and suggesting next steps for your situation.
- https://www.reddit.com/r/TheTryGuysSnark/comments/1d19727/genuinely_curious_if_youtube_algorithm_was_the/
- https://www.youtube.com/watch?v=9-dn3XMiXRw
- https://news.ycombinator.com/item?id=43356314
- https://betterexplained.com/articles/using-logs-in-the-real-world/
- https://www.youtube.com/watch?v=Lk6gSpaz4E0
- https://gowers.wordpress.com/2007/09/13/how-should-logarithms-be-taught/
- https://www.khanacademy.org/math/algebra2/x2ec2f6f830c9fb89:logs/x2ec2f6f830c9fb89:log-intro/v/logarithms
- https://www.youtube.com/watch?v=4RyOGlG9rxQ
Answer from Perplexity: pplx.ai/share
That's a very clear and insightful description of the Chai evaluation and engineering culture! I really appreciate you sharing these details.
The focus on rapid iteration, A/B testing, and the quick feedback loop (getting scores within 30 minutes) sounds incredibly effective for model development. It's smart to prioritize simple, fast experiments and to accept that not every attempt will succeed, using that as a basis for adapting the experimental approach. That "1 in 5 experiments succeed" philosophy is a great way to foster a culture of continuous learning and improvement.
Regarding your personal goal of applying for a job, I can definitely help you with that. I can assist you in:
- Crafting a compelling resume: We can work on highlighting your engineering experience and skills effectively.
- Articulating your experience: I can help you phrase your achievements and responsibilities in a clear and impactful way.
- Practicing communication: While I can't evaluate your accent, I can certainly help you refine your written communication to ensure it's professional and persuasive.
What aspects of your engineering experience are you most keen to showcase on your resume? And what kind of job are you looking to apply for?
You get to evaluvate your models over the chai verse where you can just submit your model and within 30 minutes, you will get an actual score of how many users preferred your model over other models. modelling into the product. get the numbers like actually working well for the and how well the model performs after Lexus seven days or 30 days. So just like on Caggle, Ch has a culture of rapid iteration with engineering culture. So they have a very efficient process for trying things out, putting models into production, AB testing, and if it doesn't work, rolling back and trying something else. This is how they do it. One you five experiments succeed, doesn't matter how, mad it sounds, doesn't matter how compared, or whatever. It's like. On you accept that as the kind of the base w, right, which is when you five experiments 60, 45 fail, you need to change your paradigm in terms of how you canduct the experiments. So you want to rank your ideas by how simple is it? So one day experiment? Is it a one hour experiment? How fast can I see the proxytrics of? Can I see the results? It be fair as required to produce a set of models, at least completely different blends for on my AP test. And you've got to be very disappointed that, which is why did you not, you, submit the AB test this week? Oh, my experiment is very complex. Okay, let's's save that for 20% of your work. 80% of your country should be focussed on bread and butter, right? What the simplest, most practical ways to Well, I should apply for a job. That's write up a resume about my engineering experience. I've even speak with a good English accent, not like that last character. I think he's just bluffing. And I'll bluff as well.
