SoftBank Is Betting Billions That OpenAI's Agents Will Automate Work | Hacker News Reader

SoftBank Is Betting Billions That OpenAI's Agents Will Automate Work

gizmodo.com

15 pointsrntn10 comments

There should really be bets placed on the new work that AI will supposedly open up. These investments in productivity, insofar as they reduce the workforce needed for today's jobs, just reduce the amount of money circulating in the consumer economy: smaller/fewer paychecks aren't yet being offset by lower prices for anything. Companies can on average still be more profitable with AI stuff (when it works) just conducting business with each other.

I'd like to see where the middle-class jobs are that arise from widespread deployment of AI. Who's investing in those?

Technological advancements have enhanced production efficiency, reducing the labor required for merchandise production. However, overall social demand grows slowly (an individual only has 24 hours a day, eats three meals daily, and replaces one phone annually). Consequently, more labor is being squeezed into service industries to become appendages in the value chain, or entirely excluded from production relations.

This is a self-destructive vicious cycle doomed to collapse.

> just reduce the amount of money circulating in the consumer economy

This is why we have Keynesianism, fiscal overspending, and inflation.

> where the middle-class jobs are that arise from widespread deployment of AI, Who's investing in those?

At present, truly substantial AI investments are focused on upgrading existing industries rather than pioneering new ones(besides AI industry itself)—such as medical imaging diagnostics or industrial robotics. While the steam engine led to the global expansion of railways and the internet digitized all business operations, AI has not yet to demonstrate such revolutionary transformative power.

In my book, anything that improves the quality of the output that a person can produce, or decreases the time they spend doing falls into the bucket of adding middle-class jobs. (When things get cheaper, we end up using more of them -- See Jevon's Paradox.)

Of course, there are lots of companies trying to eliminate job positions because they think AI can do it. That won't work at all (and if it does, the result will no longer be competitive than companies that use AI to help their workers do better), so those companies will be ripe for disruption next year.

You're reading this comment on HN, so you can probably figure out the next chapter of this story on your own.

Jevon's Paradox, as I understand it, applies in some domains and not others. Recently discussed here in fact [https://news.ycombinator.com/item?id=42906919] Companies disrupting each other isn't really the point. They can go on disrupting each other. What's missing are the uses that will grow employment somewhere to offset the employment that's trimmed by efficiency gains. Growing by equivalent or better wages and by the same number of people, that is. A few highly-comped individuals and many McJobs that are farther behind in real wages than they all are today is not a net plus.

>Of course, there are lots of companies trying to eliminate job positions because they think AI can do it. That won't work at all (and if it does, the result will no longer be competitive than companies that use AI to help their workers do better), so those companies will be ripe for disruption next year.

Could you please expand upon this? One of the visions I have of the mid-term future is that a company could be run entirely by a solo entrepreneur or a small executive team, with all the rest done by various AI agents. I would appreciate a good argument for why a company like that would necessarily be ripe for disruption.

To those of us that use and understand this technology, it's laughable to think that AI can replace an actual human, but I've seen enough people unable to detect obviously reckless AI recommendations floated to leadership that I'm afraid of the near future.

>laughable to think that AI can replace an actual human

Any particular human competency that you believe that AI wouldn't attain in a decade?

Figuring out how to use the AI and what to use the AI for.

This made me smile but I was thinking more along the lines of honour, conscientiousness, consideration, truth, knowing with conviction... common sense/etc. There are areas where this stuff is not appropriate but there is an unwillingness to hear that.

Well, if that's the limitation, then it exactly fits the vision I mentioned earlier, of a company that could be run entirely by a solo entrepreneur or a small executive team.

Reply on news.ycombinator.com