The race for the most advanced artificial intelligence is in full swing, and two of the leading contenders, Meta and OpenAI, are about to make a bold move. With the upcoming release of llama 3 e GPT-5, promise to take AI to a level never seen before, endowing it with human-like cognition, reasoning and memory. Are we ready for this?
A leap in quality for chatbots
Joelle Pineau, vice president of AI research at Meta, stressed that the company is working hard to ensure that these models do not just converse, but are able to reason, plan and have memory.
In the same way, Brad Lightcap, COO of OpenAI, said to the Financial Times that the next version of GPT will show progress in solving complex queries through reasoning.
A sort of "back and forth" that gives a good idea of how much the level of competition is (still) about to rise. And I didn't mention Google with his Gemini.
Towards an AI with increasingly “human” cognition
The goal of these companies seems to be to introduce typically human elements such as reasoning and "thinking" into AI. Lightcap said that we will see AI capable of tackling increasingly complex tasks in a sophisticated way, and that we are only beginning to scratch the surface of these models' reasoning capabilities.
It excites and disturbs me to think of a chatbot capable of "acting" with cognition and memory. Tools like midjourney e Sora have already demonstrated the excellence of artificial intelligence in terms of output quality, Claude e Chat GPT are great examples of how useful text-based bots can be in everyday life. Having a "thinking" instrument, however, promises a real earthquake.
An uncertain future
With so many ethical and moral questions still unresolved regarding the tools currently available, it is complicated to imagine what could be done with even more “human-like” AI models.
An official launch date for Meta Llama 3 or OpenAI GPT-5 has not yet been announced, but we can assume that the models will make their appearance during the next quarter. Maybe even earlier: Nick CleggMeta's president of global affairs, said the company hopes to begin shipping its new suite of next-generation foundation templates, Llama 3, within the next month or less.
The advent of AI models with “human cognition” undoubtedly represents a significant step forward in the field of artificial intelligence. At the same time, it is a new test for society, especially in Europe (cradle of first law which regulates the use of artificial intelligence).
Only through open debate and careful regulation will we be able to fully exploit the potential of these technologies, while minimizing the risks and pitfalls. The future of AI is as promising as it is uncertain, and it is up to us to shape the path we take.