We are not there, yet, but if AI could replace a sizable amount of workers, the economic system will be put to a very hard test. Moreover, companies could be less willing to pay for services that their internal AIs can handle or build from scratch.
There will be fewer very large companies in terms of human size. There will be many more companies that are much smaller because you don't need as many workers to do the same job.
Instead of needing 1000 engineers to build a new product, you'll need 100 now. Those 900 engineers will be working for 9 new companies that weren't viable before because the cost was too big but is now viable. IE. those 9 new companies could never be profitable if it required 1000 engineers each but can totally sustain itself with 100 engineers each.
We aren't even close to that yet. The argument is an appeal to novelty, fallacy of progress, linear thinking, etc.
LLMs aren't solving NLU. They are mimicking a solution. They definitely aren't solving artificial general intelligence.
They are good language generators, okay search engines, and good pattern matchers (enabled by previous art).
Language by itself isn't intelligence. However, plenty of language exists that can be analyzed and reconstructed in patterns to mimic intelligence (utilizing the original agents' own intelligence (centuries of human authors) and the filter agents' own intelligence (decades of human sentiment on good vs bad takes)).
Multimodality only takes you so far, and you need a lot of "modes" to disguise your pattern matcher as an intelligent agent.
But be impressed! Let the people getting rich off of you being impressed massage you into believing the future holds things it may not.
Maybe, or 300 of those engineers will be working for 3 new companies while the other 600 struggle to find gainful employment, even after taking large pay cuts, as their skillsets are replaced rather than augmented. It’s way too early to call afaict
Because it's so easy to make new software and sell it using AI, 6 of those 600 people who are unemployed will have ideas that require 100 engineers each to make. They will build a prototype, get funding, and hire 99 engineers each.
There are also plenty of ideas that aren't profitable with 2 salaries but is with 1. Many will be able to make those ideas happen with AI helping.
The more software AI can write, the more of a commodity software will become, and the harder the value of software will tank. It's not magic.
Total size of the software industry will still increase.
Today, a car repairshop might have a need for a custom software that will make their operations 20% more efficient. But they don't have nearly enough money to hire a software engineer to build it for them. With AI, it might be worth it for an engineer to actually do it.
Plenty of little examples like that where people/businesses have custom needs for software but the value isn't high enough.
this seems pretty unlikely to me. I am not sure I have seen any non-digital business desire anything more custom than "a slightly better spreadsheet". Like, sure I can imagine a desire for something along the lines of "jailbroken vw scanner" but I think you are grossly overestimating how much software impacts a regular business's efficiency
As an alternative perspective, if this hypothetical MCP future materializes and the repair shop could ask Gemini to contact all the vendors, find the part that's actually in stock, preferably within 25 miles, sort by price, order it, and (if we're really going out on a limb) get a Waymo to go pick it up, it will free up the tradeperson to do what they're skilled at doing
For comparison to how things are today:
- contacting vendors requires using the telephone, sitting on hold, talking to a person, possibly navigating the phone tree to reach the parts department
- it would need to understand redirection, so if call #1 says "not us, but Jimmy over at Foo Parts has it"
- finding the part requires understanding the difference between the actual part and an OEM compatible one
- ordering it would require finding the payment options they accept that intersect with those the caller has access to, which could include an existing account (p.o. or store credit)
- ordering it would require understanding "ok, it'll be ready in 30 minutes" or "it's on the shelf right now" type nuance
Now, all of those things are maybe achievable today, with the small asterisk that hallucinations are fatal to a process that needs to work
exactly. have you seen App Store recently? over-saturaded with junk apps. try to sell something these days. it is notoriously hard to make any money there.
Similarly flawed arguments could be made about how steam shovels would create unemployment in the construction sector. Technology as well as worker specialization increases our overall productivity. AI doomerism is another variation of Neoluddite thought. Typically it is framed within a zero-sum view of the economy. It is often accompanied by Malthusian scarcity doom. Appeals to authoritarian top-down economic management usually follow from there.
Technological advances have consistently unlocked new, more specialized and economically productive roles for humans. You're absolutely right about lowering costs, but headcounts might shift to new roles rather than reducing overall.
I am not sure it will scale like that... every company needs a competitive advantage in the market to stay solvent, the people may scale but what makes each company unique won't.
if these small companies are all just fronts on the prompts (a "feature" if you will) of the large ai companies, why do the large ai companies not just add that feature and eat the little guy's lunch?
Instead of needing 1000 engineers to build a new product, you'll need 100 now. Those 900 engineers will be working for 9 new companies that weren't viable before because the cost was too big but is now viable. IE. those 9 new companies could never be profitable if it required 1000 engineers each but can totally sustain itself with 100 engineers each.