Sam Altman recently gave a talk at Y Combinator.
He told the founders to assume that GPT5 was around the corner and that artificial general intelligence would come shortly after.
AGI is quite a hyped up term, often even used in memes.
It’s hard to know if Sam truly believes this or this is one of those mythical tech founder phrases like when Musk said self driving would be achieved in 2016.
Either way, it poses quite an existential question to most founders.
How do I build an AGI proof business?
I think this is something we will have to further unpack as we go.
Some clear options where GPT4 still seems very far off from are:
- Anything physical — manufacturing, engineering, Even logistics or services — this could be advanced by the mixture of AI and robotics and some promising results have been witnessed. But ultimately, I also think this is limited because of the lack of training data on the certain tasks compared to the structured and aggregated data of the Internet.
- Private and unique data sets (we’re available training data lacks) — private, and custom data set will become much more valuable if other AI models cannot access it and require it for certain tasks. This becomes difficult, though depending on the outcome of the New York Times lawsuit, which would render any type of data even copyrighted up for grabs for AI training.
