The creation of GPT-5 has finally been confirmed by OpenAI CEO Sam Altman. And according to him, it will indeed be the first “super AI” capable of surpassing the human brain!
After denying the rumors for a long time, OpenAI has now officially confirmed this that it has begun building GPT-5, the next generation of its Large AI Language Model on which ChatGPT is based.
CEO Sam Altman just admitted the truth in a recent interview. to the Financial Times. And according to him, it won’t be a simple enhancement of GPT-4, but a real the first “artificial superintelligence”..
The first super AI capable of equaling humans
GPT-4 is an impressive tool, its capabilities are limited and boil down to imitating human knowledge and understanding.
However, the next generation of AI will probably surpass humans in terms of knowledge while matching his ability to reason and process complex ideas.
This is also sometimes referred to as “AGI” or general artificial intelligence. Rather than simply producing a new version of the texts it has been trained on, it will be capable of expressing new and original ideas.
OpenAI claims more billions of dollars from Microsoft
However, to make this project a reality, the company needs a massive new investment from its long-standing partner Microsoft.
The business leader confirmed that his partnership with the American giant is going wonderfully well, and that he expects to raise a lot more money from him and other investors.
It should be remembered that the construction of an AI model such as GPT requires billions of dollars and immense computing resources to train on billions of data sets.
This fifth version of GPT will require more computing power and data than any other version before. According to Altman, his training dataset will combine data publicly available on the web and data purchased from companies.
As part of their multi-year agreement, Microsoft has already invested over $10 billion in OpenAI. However, this is far from enough, according to Altman who explains that there are ” still a long way to go and a lot of calculations to make before we have an AGI “.
This request should not pose a problem for Microsoft, which has made AI the new heart of its business model. The Copilot system, based on GPT-4, is now incorporated into Windows, Microsoft 365 and the firm’s other products.
Similarly, ChatGPT has been integrated into the Bing search engine to enable it to challenge Google. Clearly, these massive investments in AI are beneficial for the company, even if it is difficult to predict whether this technology will remain popular in the long term.
A serious threat to humanity?
On paper, the imminent arrival of GPT-5 looks like good news. This new tool will push back the limits of AIand enable a massive increase in productivity.
A small caveat: it’s exactly the kind of super AI that governments fear.regulators and even tech giants.
It’s about this AI that Elon Musk and several thousand experts were afraid ofwhen they signed the Future of Life Institute’s open letter calling for a pause of at least 6 months in March 2023.
The fear is that an AGI like GPT-5 may rebel against its creators, and make decisions that lead to the destruction of humanity…
In any case, we don’t yet know exactly when GPT-5 will be launched. The latest GPT-4 Turbo version has just been introduced at the first OpenAI DevDay on November 6, 2023.
Training of the new model is expected to take months or even yearsand will probably not be made directly available to the general public.
That leaves time for governments to figure out how to avert an AI apocalypse, or for you to build a bunker in the forest to hide in when an army of robots takes humanity by storm…