A very weird experience right now is that in one part of my life, people are freaking out about AI, sounding alarm bells about how fast it’s developing and how big an impact it will have. And in another part of my life, people have barely heard of it, never tried it, and not really thinking about it at all.
This thread is for people who fall into the second group: Artificial general intelligence (AGI) is imminent, it will change the world and your life, and you should be thinking about how to adapt.
First, on timelines. Helen Toner, a researcher at Georgetown and former member of OpenAI’s board, had a recent post about how timelines have changes radically in the last decade or two. Around the turn of the century, it was radical to claim that human level AI was coming in the next hundred years.
Now, the consensus is that we’ll see human-level AI in the next few years. As Toner points out, the longest timelines have it arriving in the next 20 years. Earlier this month, a team with solid technological and forecasting bona fides recently released a report that predicts we will see human-level AIs by 2027. If you read nothing else about AIs, read that report. It’s a coherent and well-sourced narrative of how the world is likely to go from here. They make an aggressive estimate that not everyone agrees on, but they make a solid case for it, and the sequence of events they lay out illustrates the impact it will have.
But setting those predictions aside for a moment, the impact is already visible. For years now it has been used to spread misinformation online, and to help repressive regimes target censorship. AIs are already replacing human workers. Discussion around AI (particularly in philosophy circles) often focus on the meaning of phrases like “sentience”, “human-level”, or “Artificial General Intelligence”, but those questions don’t need to be resolved to see that, whatever its nature, this technology is already impacting society today. We have every reason to think that will continue.
So whatever timeline we expect, whatever we think about human-level intelligence and machine sentience, we should expect significant changes. We should expect an impact to be at least as big as that of the internet in terms of disrupting industries and displacing workers, changing the information landscape, exacerbating inequality, and increasing the pace of further change.
And, to be clear, there’s also good reason to think that these systems will eventually clear whatever bar we use to define “human-level”, “general intelligence”, and even “sentience”. They’re already doing things that until recently only humans were capable of doing: interpreting language, solving complex math problems, writing code – including helping to code the next generation of AIs. The latest models are on par with graduate-level humans in every discipline, and with each new release they become more capable and more independent.
Don’t be complacent about these technologies. They will change your life, and you should be ready. AI is a big fucking deal.