Brace yourselves for another round of hype and hysteria mixed for us by OpenAI that will happen this fall. In another great article, Ignacio de Gregorio sums up what to expect from a new generation of OpenAI models and marketing materials as their Chief Hype Officers truly mastered the art of getting people's attention to promote company products. Everything indicates that we will get a model (models?) trained on a large dataset of reasoning task examples and with reward models - that will imitate thinking like a seasoned illusionist. We will get another iteration of the "intelligent database" that skillfully imitates raw intelligence (which requires active search / active reasoning instead of retrieval). It will impress the users with better capabilities than anything on the market today, and disappoint the practitioners with failing miserably on complex tasks that can't be retrieved from the model's associative memory. Link to the article in the comments 👇
Vidar Hokstad
3mo
I'm always amused at attempts to downplay the abilities of LLMs by describing them this way, as almost invariably the descriptions downplay how little the descriptions say about limits to actual capabilities, and computability: An associative memory + IO + a loop is trivially Turing complete. Absent evidence the Church-Turing thesis is wrong, the reasonable assumption is that with the right input they can compute anything a brain can. Without addressing the actual *models*, this criticism is meaningless - the *models* may (and likely will) limit them, but the computational power of the underlying architecture does not.
I am really excited to see what new capabilities the new model will hold, Maciej Wolski. What is your biggest hope?
Maheep Gupta
3mo
BTW, I really like this bit - "that will imitate thinking like a seasoned illusionist."
Maheep Gupta
3mo
It will prolong the hype for another few months yet won't really contribute in many meaningful ways to a production-ready usable software. Getting people's attention. that is pretty much it.
Sandijs Aploks
3mo
Finally, more and more specialists have the courage to openly speak about the naked king 🤣 I am not 100% sure, but Google folks were one of the first who explained why LLM is not the AI we expect. Still, the mass hysteria began, and the Google had to release hallucinations engine too.
Julius Domingo
3mo
Hoping the consumer price will go down as well
Great article, thank you! Looks like a totally reasonable path for OpenAI, a market leader in a fragile position between setting new standards, being closely followed by competition, and not only balancing on the cash and profitability line, but also on the ultimate borderline in the form of energy supply to cover the computation requirements.
Emanuel Piza
3mo
Looking forward to it!
Mikus Krams
3mo
Disappointed practitioners is the name of the game.
Human-like Agents | Closing the gap between AI and REAL Intelligence
3mohttps://medium.com/@ignacio.de.gregorio.noblejas/thorough-analysis-of-openais-new-leaked-strategy-a3035381850c