As 2022 draws to a close, the chat AI based on GPT-3 (Generative Pre-trained Transformer 3) has been taking the internet by storm, with millions of users beginning to ask it questions. Is humanity on the way to birthing a true artificial general intelligence (AGI)? I asked GPT-3 that particular question, and this was the answer:

It is difficult to say for certain whether or not humanity is on the way to creating a true artificial general intelligence, as there is no clear consensus on what exactly a true AGI would be or what it would be capable of. However, it is clear that significant progress has been made in the field of artificial intelligence, and many experts believe that it is only a matter of time before we see AGI-level capabilities in machines. That being said, it is also important to note that the development of AGI is a highly complex and uncertain field, and it is impossible to predict exactly when, or even if, it will be achieved.

Whether GPT-3 is truly intelligent, it’s clearly just the beginning. At least according to this week’s returning guest on Unsupervised Learning, Nikolai Yakovenko of Deep NFT Value, a machine-learning engineer who just last month took time to discuss what was going on at his old employer, Twitter. Yakovenko is not entirely surprised by GPT-3’s abilities, though he admits progress is faster than expected. He tells Razib that there are more advanced versions of GPT-3, and Americans must brace themselves for a new technological revolution. He believes Westerners, in particular, due to their religious and metaphysical frame are unprepared for the changes that AI may produce in our society. Yakovenko rejects the most dystopian and negative predictions of hostile AI and singularity and argues that the Japanese are a model of a culture that is more sanguine about the emergence of human-like machine intelligence.

Razib and Yakovenko also discuss the utility of AI-generated art and how it might replace or complement human artists.

Nikolai Yakovenko: GPT-3 and the rise of the thinking machines

Razib Khan's Unsupervised Learning