© Unsplash/Levart_Photographer
This is the absolute Holy Grail in the world of Tech in 2024: artificial general intelligence (AGI). Behind this term hides the idea that an AI could be capable of outperforming humans in certain tasks and of reasoning for itself. Such an innovation would bring in colossal sums of money for the company which would be ahead of its rivals, so much so that a race against time is launched. This week has been rich in announcements and promises from this point of view.
Taking AI to a new milestone
Meta and OpenAI have in fact communicated on this subject through their managers. Mark Zuckerberg's firm plans to deploy Llama 3 in the coming weeks. For their part, ChatGPT developers stressed that their next GPT-5 language model will be released soon.
Le Financial Times< /em>was able to speak with Brad Lightcap, COO of OpenAI, who promises that this new product will show progress in solving “difficult problems” such as reasoning.
Our colleagues also report the remarks made by Yann LeCun, Meta's scientific manager in terms of AI this Tuesday during a conference in London. He evokes current generative AIs which “produce one word after another, really without thinking or planning” and which commit in fact still “stupid errors”.
If these tools become capable of reasoning, they will be able to search for possible answers and plan sequences of actions, explains the manager. He concludes by affirming that it is the “large missing piece we are working on to enable machines to reach the next level of intelligence”.
According to the British media, Meta could benefit from these advances by introducing AI functionalities in WhatsApp as well as in its Ray-Ban smart glasses. For its part, OpenAI prefers not to say more for the moment, but teases future announcements.
Experts call for caution
In any case, this AGI theme is indeed on everyone's minds. Elon Musk recently addressed the latter in a podcast. The entrepreneur even ventured a prediction: “I think we will have an AI smarter than any human probably towards the end of next year”.
Nvidia CEO Jensen Huang is a little less optimistic, saying it will take another five years to get there. Finally, let us point out that some renowned experts on AI say they are worried about seeing such technology see the light of day. This is particularly the case of Yoshua Bengio and Geoffrey Hinton who invite companies to think carefully about the consequences of such innovations on human beings.
📍 To not miss any news from Presse-citron, follow us on Google News and WhatsApp.
[ ]