Capital Account: More people turn to ChatGPT and other large language models for answers, but they don’t add to the stock of knowledge https://t.co/quYCxeH398 ...

I don't think Sam Alden has done a service to the world by talking about how close AGI is. I think he has made several predictions now that are wrong and that were obviously wrong at the time he made them. I think AI will probably lead to the end of the world. You know, he's made illusions to things. He did a world tour where he spoke to every major leader the world over to tell them, hey, this technology is going to poses an existential threat. And I think that was academically disingenuous and I think did ...

O să mă folosesc de Eminem, care a spus: "Snapck to reality, here comes gravity". Încerc să vă aduc un pic înapoi cu picioarele pe pământ, deși cred că nu e rău să visez. Până în ultima clipă m am gândit să schimb ceea ce voiam să vă povestesc cu totul, pentru că am avut o discuție scurtă, dar intensă pe hol înainte de a intra aici și cred că puțini dintre voi aveți habar ce se întâmplă în spatele scenei, ce se întâmplă în lumea care ne înconjoară și decuplarea asta de la de la comunitate și de la o voce co ...

Large Language Models Evolution - Scaling laws 表明，增加计算量、数据和参数可以提高 Transformer 模型的性能，并推广到其他领域 [2][3] - 随着模型规模的扩大，性能持续提高，并在中等数学难题的解决率上有所体现，尤其是在提示模型展示思维链时 [5][7] - 通过强化学习和人类反馈，模型能够更好地遵循指令，从而实现聊天机器人等应用 [10][11] Inference Time Optimization - 通过生成多个响应并进行多数投票（自洽性），可以在推理时提高性能 [15] - 顺序修改之前的响应，特别是在可以验证答案的领域（如数学和编程），可以显著提高性能 [16][17] - 在可以验证答案的领域，推理时间计算的扩展可以转化为智能 [19] Reinforcement Learning for Autonomous Coding - 强化学习是下一个扩展前沿，特别是在可以自动验证输出的领域 [24] - 经验时代将通过强化学习构建超级智能系统，尤其是在具有自动验证的领域 [25] - 自动编码是一个扩展强化学习的绝佳领域，因为它具有验证输出的能力 [30][31] Challenges in Scaling Reinforcement Learning - 扩展强化学习比扩展 LLM 更具挑战性，因为它需要多个模型副本以及训练和推理循环 [29] - 在强化学习中，奖励模型的奖励函数设计是一个挑战 [29][30] Reflection's Mission - Reflection 致力于构建超级智能，并以自主编码作为根本问题 [33] - Reflection 团队由在 LLM 和强化学习领域有开创性工作的 35 位先驱组成 [33]

Talent Acquisition & Competition - Meta is attempting to recruit OpenAI employees, but OpenAI's CEO downplays the significance, suggesting Meta is not acquiring top-tier talent [1][2] - OpenAI believes its shares offer greater potential upside compared to the publicly traded Meta [2] - Meta is offering substantial multi-year deals, reportedly worth $300 million, to attract scientists and engineers, with some receiving $100 million upfront [3] Cultural & Performance Implications - OpenAI raises concerns about the potential cultural impact within Meta due to high, guaranteed compensation packages, questioning the work ethic of new hires [4] - The industry observes that employees attracted by high paychecks may lack company loyalty and could easily leave for better offers [5] Large Language Model (LLM) Landscape - Meta is actively working to enhance its Llama LLM, but it is currently considered less successful than those developed by Anthropic, OpenAI, and Google [6]

Can Meta deliver on its AI. Got to the right camera in the end there. Lot of all four to get there.Uh, can they deliver on their AI ambitions as the competition heats up. Joining us now, New York University professor emeritus and AI expert Gary Marcus. Gary, great to have you have you with us.I maybe maybe this is a tough question to to answer, but can you buy in outside AI expertise and incorporate it quickly, easily, or is it as complicated as trying to do a massive corporate merger and the kind of teethi ...