The technological startup OpenAI and the social media company Meta will soon launch new artificial intelligence (AI) models that will be capable of reasoning and planning, key steps towards achieving in machines a cognition analogous or superior to that of humans. Executives from OpenAI and Meta have indicated that they are preparing the launch of the next versions of complex language models, the systems that power generative AI applications like ChatGPT.
Meta has indicated that it will begin rolling out ‘Llama 3’ in the coming weeks, while OpenAI, backed by Microsoft, has indicated that its next model, which is expected to be called ‘GPT-5’, will arrive “soon”. In this regard, Joëlle Pinau, Vice President of AI Research at Meta, has indicated that the company is “working hard to make these models not only speak, but also reason, plan… have memory.”
Meanwhile, Brad Lightcap, COO of OpenAI, has stated that the next generation of GPT will show advances in solving “hard problems” like reasoning. He mentioned this in an interview with Financial Times, where he said we will start to “see AI capable of taking on more complex tasks in a sophisticated way”. “I think we are just starting to scratch the surface of the capacity these models have for reasoning,” he added. Lightcap also argued that current AI systems are “really good at small specific tasks”, but that their capabilities are still “quite limited”.
