Kyle Corbitt

When people ask why—despite all the risks—I'm optimistic about our AI-dominated future, I sometimes struggle to paint a legible picture of what the good version of that future actually looks like to me.

@UnderwaterBepis Thanks for reading!

I liked this, some minor quibbles (AI Alignment *must* be transparent and auditable by anyone or it has failed) but its worth reading. I especially appreciated this last bit

@kavinstewart Yes, let's build it! 💪

@corbtt Loved it. A future to aspire to

When people ask why—despite all the risks—I'm optimistic about our AI-dominated future, I sometimes struggle to paint a legible picture of what the good version of that future actually looks like to me.

@rihim_s Thanks!

@corbtt that was an amazing read

@sethbannon Agreed. Saw this short story going around this is super positive and optimistic, but if you dig into it—the big, hard questions are being asked and answered thoughtfully: x.com/corbtt/status/… Very positive outcomes are possible, but that doesn't mean they come for free.

@EvanOwen @sethbannon Thanks for the kind words!

These are real concerns from real people. Truly important questions to get right. It's refreshing that they're addressing them and it gives me hope we'll solve them. Feels like acknowledging the hard road ahead while staying positive about our ability to build a brighter future.

@arb8020 Thanks for sharing!

recommended read

@KyleMcnease Thanks for reading!

If only the paper was titled, “Love Is All You Need.” That we are loved…this is what makes us special. It is not our capacities.

if you love ai, you have to read this. if you don't love ai, you really have to read it. it's a great narrative about a wonderful future that may await us. thank you for sharing @corbtt!

@bradhilton Thanks for the kind words!

i keep coming back to this. it's a story about superintelligence taking over but where alignment is solved and everything basically ends up fine. compelling, sober, reasonable writing from an expert deep in the weeds. makes a lot of the other speculative scenarios feel silly.

@willccbb Thank you Will!

@aidan_mclau you would enjoy this piece

This is a great read about a lovely future! I wish there were more ideas of good outcomes available for us to think about like this

First, make it exist Then, you can build it Or something like that

@willccbb @aidan_mclau thanks for sharing @willccbb. @aidan_mclau I'd love to get your take on this future if you get the chance to read it!

the most radical belief someone with short timelines can have is that things will go normally

@infornomics Thanks for sharing! ❤️

@TheRyanHartman Thanks for reading and sharing!

@adi_baradwaj An AI could become superhumanly persuasive though subtly controlling someone's whole media environment and tailoring its content/arguments to them. It would have tools no cult leader or politician has had.

There's a recurring pattern that I keep seeing in work coming from the apocalyptic branch of the AI safety community, and it goes something like this: 1) Pick some intellectual "talent" that might be attributed to a person (e.g. persuasiveness, discernment, charisma, etc.) 2) Model this "talent" as a scalar quantity that is primarily determined by factors endogenous to the individual, as opposed to environmental/situational factors 3) Assume that this scalar quantity can be made arbitrarily large 4) Use this model to make predictions about the future of AI You can see this here with AI 2040's insinuation that "superhuman persuasiveness" is an idea we should be taking seriously It's not obvious to me at all that "persuasiveness" is a human talent, as opposed to a sociological random process that we retroactively perceive as a human "talent" To be clear, certainly it's true that a star debater might be marginally more "persuasive" than someone who's not! But I don't think a cult leader or a popular politician is 1,000x or 1,000,000x more "persuasive" than an ordinary person Rather, they're perceived being "persuasive" because they happen to be the figureheads for a complex sociological preference cascade. Their "persuasiveness" isn't really a thing you can causally influence at the individual level, and definitely not in an unbounded way In general, I think a lot of the AI 2040-style forecasting work does a poor job of dealing with this kind of irreducible complexity inherent to the universe. They usually just like to pretend it doesn't exist Not a huge fan of this pattern

@sebkrier I really appreciate concrete visions like this. I tried to answer a similar question through a short story set in a near-future post-AGI world, exploring what a good life looks like under those constraints. Would love to get your take on it! x.com/corbtt/status/…

Positive visions of the future are easy to produce, so it's understandable that people dismiss them. I think that's a mistake, and that disempowerment-fatalism is just as easy. If I had to imagine a nice 'post-AGI' vision, it would look like this: I have to work much less, but still have a chunk of the year working if I want to. There is no 'work day' but there are 'work hours'. Opting out is sustainable for those who don't want it. A Venetian-like economy creates new roles, tasks, and quests that people never imagined could exist (I think full substitution will take much longer to bite than people assume). The rest of the year is a mix of travelling, helping design a new city somewhere, doing research/writing that is more longitudinal in nature, and making music with new synths designed at some workshop in a different continent (the world of atoms is flourishing, and globalization is so back). Low goods/services prices let me and my friends coordinate a few impromptu hangouts abroad every now and then - we realize it's much easier to nurture social life when time is not a blocker, and coordination costs are lowered by agents. My family's health issues are managed, giving them more years to live (clinical trials have sped up significantly). The decreased dependence on living in a large capital city combined with the affordability of building transport infrastructure leads to a flourishing of new towns in places that would otherwise languish; people are excited to build them (yes building is legal again). There are a number of large scale projects to create colonies in space for the more ambitious pioneers, and many highly paid jobs there too. Coordination tech lets people self-select into hubs, communities, towns that fit their vibes. In fact, the 'age of polarization' is now behind and a lot of governance is pushed down to the local level, which lowers the stakes of national identity fights. Conflicts naturally remain, but the 'boring majority' is no longer faced with having to choose between two flavours of outrage and righteousness. Government is human-led and agent-mediated. Automated dispute resolution lets human courts deal with more important and significant cases. Citizens can automatically simulate the likely impacts of proposed laws, and agents help demystify them. The machinery of government still has humans, but it's far leaner; there's a neat separation between instrumental tasks (automated) and normative ones (augmented). Politicians are rewarded more, but they're also under automated scrutiny by constituents - e.g. commitments are easier to monitor and enforce. I could go on! But as I write this, some will rightly say 'well this just sounds like a nice sci-fi story', and tbh they wouldn't be entirely wrong. For each sentence I can find a story why it might not happen. It's still helpful to outline a positive vision of the future to have an idea of what to aim for, but ultimately much of this will come down to resourceful political entrepreneurs, founders of new companies, important changes in legislation and culture and so on. You don'tt get any of this automatically, it's an endogenous dynamic that depends on how many people exercise their agency. There is of course a degree of determinism to technology, and wider dynamics that cannot be stopped - but people too often point to them as if disempowerment is the only possible outcome. I think that's 'spectator cope', and a lot of shaping will remain possible. Societal change is inherently hard and diffuse, but it's certainly not impossible - too many people despair at being unable to control wide structural changes, but that's missing the trees for the forest, in the most literal sense.

@jrhwood Totally agree. I wrote up a more positive (but realistic) vision that I would love to get your take on! x.com/corbtt/status/…

Much of EA scripture relies on dystopian sci-fi tropes, we need to paint a positive vision for ai.