Transformative AI (TAI) is AI that fundamentally changes the economy, increasing productivity at an unprecedented rate and likely completely changing (if not eliminating) human work and companies as we know them.
Whenever I read these 10x gains (in speed, not quality) or the idea that we will leave the average worker behind, I have to ask why.
What, exactly, is the benefit of making an app or a branded web experience or an e-commerce site, a thousand custom social ads if very few people have jobs left with which to purchase your product or service?
And what is lost when we focus on speed gains? Well, my peace of mind, for one. I can generate a mountain of writing or visuals, but I still need human spans of time to analyze, decide what is useful, and time to try things that don’t ultimately work or meet the needs of the project. The fallacy that I’m saving time using AI means I can’t do the very role you describe: be innovative, be analytical. I just become a reverse-centaur.
The fallacy that AI has hypercharged my productivity (plus the uselessness of merely being faster), has led my firm to lay off wildly talented, innovative people, as well as juniors. They’re chasing the gains their investors are told by articles like this are possible, but they’re doing it by doubling my workload without a clear picture of how I can achieve those gains in our current processes, with the requirements of our projects. I’m the pachinko machine. It’s on me to be the innovator but also I have less than no time to do that. It’s on me to purchase subscriptions if I want to try a new workflow.
The 10x gains are also nearly free today (aside from the terrifying environmental cost) But tomorrow, I expect these companies to turn on the money faucets once they have lock-in after firms fire all their talent.
The only disruption that would really be meaningful to me is one that disrupts the value of the owning class, and puts power back with workers to use AI to seamlessly collaborate while getting their basic administrative needs covered at a reasonable cost. But what even would we do, I wonder, and why? It may be that a disruption that large renders everything from custom apps to advertising to most business software obsolete. How, even with these miraculous 10x gains, would it be worth the effort to make something for a diminishingly small number of users or customers?
In my most hopeful thoughts, it becomes a value restructure where we return to smaller agencies (my small agency was gobbled up and pulverized by a large agency), and more creative, delightful, and heartfelt work, directed by the people who want to make it, and that the people who positioned themselves as extractors of my labor for their own enrichment are the ones who are irrelevant, not me.
I think this evades one critical aspect on the economy. How will it work if less and less people does not have an income due to lack of jobs (replaced by AI). It could also be interesting to rethink how the AI can add value to a product. As far as I know, machines don´t add value, they add cost, but in the AI scenario, they seem to do both.
I have the same question as the other commenters. How do these economic models account for who is now buying all these products and services if there’s no one left earning an income? Do we suppose that the 10,000 person company is now 1,000 10 person companies all doing vaguely similar things?
The assumption is that the economy will be *many* times bigger, with lots of new things being produced. When I say "produced", that doesn't necessarily mean manufacturing of physical products. It could equally well be service-as-software or micro-targeted content like my music videos (but much better in 10 years!).
Again, I find myself skeptical that many factors more items -- physical or not -- will be produced. I suspect we're reaching a saturation point for attention, and we're certainly over-producing most physical amenities. I believe we may make different things than we did in the past, but not much more. Again, I am left with the question of why we need more of anything, or faster of anything. I don't have the bandwidth to digest the movies, TV, music, news, micro-content, tutorials I'm interested in as it is. We can continue to flood people's feeds with more and more, but they still only have two-ish eyeballs and 16 waking hours.
Thanks Jakob. Yes, I understand the distinction in production.
But if - as hypothesised, super intelligence or TAI is an actual thing - then the marginal cost of replicating or spinning up service as a software is trivial and low value. Companies that have proprietary data sets seem to have some sort of benefit in that they have domain expertise that an LLM may not.
Where does this leave the 10 person team? Pumping out trivial micro content that no one really cares about? Have you been on SM / LinkedIn recently? It’s an awful swamp of undifferentiated nonsense that’s getting worse by the hour as everyone tries to dance to the tune of someone else’s algorithm, designed to induce industrial scale adhd.
The implicit assumption seems to be: new tech displaces old paradigm > market reconfigures after period of adjustment > new jobs and new economy emerges.
This is based largely on “well, that’s what happened last time”. I hope so.
Whenever I read these 10x gains (in speed, not quality) or the idea that we will leave the average worker behind, I have to ask why.
What, exactly, is the benefit of making an app or a branded web experience or an e-commerce site, a thousand custom social ads if very few people have jobs left with which to purchase your product or service?
And what is lost when we focus on speed gains? Well, my peace of mind, for one. I can generate a mountain of writing or visuals, but I still need human spans of time to analyze, decide what is useful, and time to try things that don’t ultimately work or meet the needs of the project. The fallacy that I’m saving time using AI means I can’t do the very role you describe: be innovative, be analytical. I just become a reverse-centaur.
The fallacy that AI has hypercharged my productivity (plus the uselessness of merely being faster), has led my firm to lay off wildly talented, innovative people, as well as juniors. They’re chasing the gains their investors are told by articles like this are possible, but they’re doing it by doubling my workload without a clear picture of how I can achieve those gains in our current processes, with the requirements of our projects. I’m the pachinko machine. It’s on me to be the innovator but also I have less than no time to do that. It’s on me to purchase subscriptions if I want to try a new workflow.
The 10x gains are also nearly free today (aside from the terrifying environmental cost) But tomorrow, I expect these companies to turn on the money faucets once they have lock-in after firms fire all their talent.
The only disruption that would really be meaningful to me is one that disrupts the value of the owning class, and puts power back with workers to use AI to seamlessly collaborate while getting their basic administrative needs covered at a reasonable cost. But what even would we do, I wonder, and why? It may be that a disruption that large renders everything from custom apps to advertising to most business software obsolete. How, even with these miraculous 10x gains, would it be worth the effort to make something for a diminishingly small number of users or customers?
In my most hopeful thoughts, it becomes a value restructure where we return to smaller agencies (my small agency was gobbled up and pulverized by a large agency), and more creative, delightful, and heartfelt work, directed by the people who want to make it, and that the people who positioned themselves as extractors of my labor for their own enrichment are the ones who are irrelevant, not me.
I think this evades one critical aspect on the economy. How will it work if less and less people does not have an income due to lack of jobs (replaced by AI). It could also be interesting to rethink how the AI can add value to a product. As far as I know, machines don´t add value, they add cost, but in the AI scenario, they seem to do both.
I have the same question as the other commenters. How do these economic models account for who is now buying all these products and services if there’s no one left earning an income? Do we suppose that the 10,000 person company is now 1,000 10 person companies all doing vaguely similar things?
The assumption is that the economy will be *many* times bigger, with lots of new things being produced. When I say "produced", that doesn't necessarily mean manufacturing of physical products. It could equally well be service-as-software or micro-targeted content like my music videos (but much better in 10 years!).
Again, I find myself skeptical that many factors more items -- physical or not -- will be produced. I suspect we're reaching a saturation point for attention, and we're certainly over-producing most physical amenities. I believe we may make different things than we did in the past, but not much more. Again, I am left with the question of why we need more of anything, or faster of anything. I don't have the bandwidth to digest the movies, TV, music, news, micro-content, tutorials I'm interested in as it is. We can continue to flood people's feeds with more and more, but they still only have two-ish eyeballs and 16 waking hours.
Thanks Jakob. Yes, I understand the distinction in production.
But if - as hypothesised, super intelligence or TAI is an actual thing - then the marginal cost of replicating or spinning up service as a software is trivial and low value. Companies that have proprietary data sets seem to have some sort of benefit in that they have domain expertise that an LLM may not.
Where does this leave the 10 person team? Pumping out trivial micro content that no one really cares about? Have you been on SM / LinkedIn recently? It’s an awful swamp of undifferentiated nonsense that’s getting worse by the hour as everyone tries to dance to the tune of someone else’s algorithm, designed to induce industrial scale adhd.
The implicit assumption seems to be: new tech displaces old paradigm > market reconfigures after period of adjustment > new jobs and new economy emerges.
This is based largely on “well, that’s what happened last time”. I hope so.