There seems to be two dominant streams in the development of AI.
The first stream seems to be rooted in its ability to accelerate a research process. This is best exemplified by the Nobel Prize in chemistry 2024[1], given to David Baker, Dennis Hassabis, and John Jumper. Hassabis and Jumper come from Google, and they presented an AI model call AlphaFold2 which helps predict the structure of all 200 million proteins that researchers have identified. It usually takes researchers anywhere from several months to years to actually discern the structure of a protein, so the actual productivity given by AI process in research is huge.
In Dario Amodie’s piece, Machines of Loving Grace[2], He did indicate that the fields that AI could have the most profound impact on, is biology, health, neuroscience and the mind. It makes sense that humanity will tackle the most pressing health issues with the ageing population of the baby boomers. The areas that AI will probably target all those that Peter Attia terms as the four horsemen of health: cardiovascular disease, cancer, cognitive impairment and diabetes. Amodie indicates that could be huge breakthroughs in these fields and there is a potential for human lifespan to increase to 150.
The second stream is for AI to be an agent. Agentic AI will act as your personal assistant. I think this is the most powerful mass market use case.
An example of this is: Let’s say you looking for a restaurant to go to with your friends tonight. You just ask the AI “oh I’m looking for Korean BBQ near me, filter with Google reviews”. The personal AI assistant comes back with several choices, you’d say “OK try the first one, can we book for 7:00”. It’ll come back with “There no reservation for 7:00 PM, seating is 6:30 PM or 8:30 PM, but the second option, which is as close to you, with kind of equivalent Google star ratings, and they have a reservation at 7:00 but if you prefer or I can book the 6.30 PM sitting at the 1st place as there is only half an hour difference”. The personal assistant AI here would be able to reason, give small pieces of advice and execute simple tasks, like booking a restaurant, texting your friends about the restaurant booking and letting them know the nearest car parks etc. I think the mass market case for AI is the trained AI assistant that will be ubiqutious in the next 5 to 10 years.
[1] https://www.nobelprize.org/prizes/chemistry/2024/press-release/