Every year, Sohan Conference Foundation organizes an ideas competition where tech experts and startup CEOs also discuss and speculate what the future will bring. On this occasion, Patrick Collison, Stripe’s manager, interviewed his OpenAI colleague Sam Altman about the risks and possibilities of artificial intelligence.
Furthermore, he revealed what the most common uses users give to his tool: “The most common use of ChatGPT is the summarization of texts to be able to keep up with email and Slack messages.”
The talk moved quickly and the first topic was economics: World Coin, Altman’s “crypto” project. In his view, AI could vastly increase real economic growth by identifying the most productive investments and designs for companies.
In this way, AI can have more relevant and deeper economic effects. He himself gives as an example that an AI that can effectively allocate funds to cancer research could save billions of dollars while accelerating progress in the search for a cure.
The keynote served quickly as to the impact of ChatGPT and what OpenAI is doing in the web search industry.
ChatGPT’s growth in the search segment. AP photo.
“Language models are going to change Internet search a lot, but they are not a threat to search. And they will only be a threat to Google if they don’t do something about it, and Google clearly isn’t going to do that.” About anything,” Altman said.
In addition, he revealed that he is developing a program that allows for natural interactions with AI agents (programs like AutoGPT, which use linguistic models autonomously, without relying on humans), which have significant There can be social consequences.
Relying only on reinforcement learning for AI is not a good long term solution. The key to achieving ‘alignment’ (the term for convergence of AI/human interests) is to understand the inner workings of the models.
“I think there are some similarities between nuclear materials and AI supercomputers and it is in this area that we can find more similarities and inspiration than others, to propose something like the IAEA for artificial intelligence”.
So he understands that, while the temptation to draw too much inspiration from past technologies should be avoided, the establishment of a global regulatory agency for powerful AI training systems is vital to safety.
AI: The sterile debate of Twitter
AI discussion on Twitter is unproductive.
When asked whether these advances are born in a society that looks at how these intelligent systems do their jobs, the OpenAI CEO accused the social network of being against the debate.
He believes that there are a lot of discussions that contribute nothing because they are not really extrapolated to the actual job that comes to a specific port. In this way, a lot gets lost along the way and they are useless debates and byzantine discussions that reach no port.
Many people who say they are concerned about the safety of AI spend days on Twitter simply saying that they are very concerned about the safety of AI.
“There are people who are very concerned and do great technical work on it, but we need more people like that. The world doesn’t need more people writing long philosophical discussions on Twitter,” Altman said.
no
