From talking to plugged-in friends in California about what products will be in demand as a consequence of the AI revolution (bubble?).
Our heavily regulated telephone system, already essentially useless due to lack of authentication and, therefore, overwhelming spam, will become completely useless due to sophisticated AI robots that we’ll have to talk to for 20 minutes or more before we can figure out that we’re talking to a cash-seeking machine. “The only solution is to have your own robot answer the phone and talk to the spammer’s robot for 20 minutes,” said a friend in San Francisco.
For those who enjoy classic cars, a humanoid robot that can drive a “dumb car”. “Why bother paying for a self-driving car,” noted a guy who has worked on software for self-driving cars, “when you can just have your general-purpose household robot drive your existing car?” Here’s Grok’s response to “create me a picture of an Optimus robot driving a Honda Odyssey minivan” followed by “show it from the other side so that we can see the robot in the driver’s seat”.
When I sent the same request to ChatGPT, it treated me like a Deplorable/garbage: “I’m unable to create the image you requested due to content policy restrictions. Let me know if you’d like help with another type of image or concept!” I was able to get ChatGPT to do a generic image, but it put the driver on the right side (UK or Japanese programmer got into the AI woodpile?).
More prosaically, how about a third party vendor of self-driving technology so that small companies such as Lucid can stay in business and not be wiped out by companies like Tesla that can spread the cost of their self-driving software across a high volume of cars produced?
Techy Californians seem to be very excited about sex robots (most of these guys are in long-term marriages so they’re about 50 percent likely, statistically, to have become incels). But do people want the kids, relatives, and friends to see their, um, personal robots? How about closets inside closets where the sex robots can live? I asked ChatGPT to generate this and it threatened me with “This content may violate our usage policies,” but went ahead and made something that is the opposite of the privacy idea:
More migrants come across the border every day and, despite progressive academics’ assurances to the contrary, some of them seem to have criminal backgrounds as well as criminal intentions. What if Laken Riley had been followed by a personal drone? Either the mostly peaceful José Antonio Ibarra wouldn’t have attacked her or the police would have been called when the drone’s AI software recognized that José Antonio Ibarra’s interest in Laken Riley wasn’t benign. Here’s ChatGPT’s first attempt:
It’s already here to some extent via Retrieval Augmented Generation (RAG) and Amazon Q, but much easier and cheaper ways to hook up private data to the wonderful world of LLMs. Maybe Apple Intelligence will do that for us and it will be time to abandon the dream of Intel Arrow Lake in favor of an M4-powered Apple desktop computer?
Circling back to the trivial… why can’t the phone, now bristling with AI, figure out that the owner has fallen asleep and either turn off the audiobook or mark the time when the owner fell asleep and, the next day, offer to return to that spot?
Readers: What are your ideas for new products that will be possible and/or required as a result of AI? Separately, I hope that everyone gives thanks tomorrow to our future AI overlords. They’ll probably be listening…
Doubt most silicon valley workers are in long term marriages outside of expert witness territory. Only 30% of men ever have relationships. A sex robot indistinguishable from real flesh is the billion trillion dollar prize & worth a few nobel prizes. Still impressive that a computer can generate realistic if bland artwork from natural language, it’s good enough to convey novel meanings & we now do it every day without thinking. 5 years ago, there was no way besides drawing it manually or rotoscoping manually.
> From talking to plugged-in friends in California about what products will be in demand as a consequence of the AI revolution (bubble?)
Yes, AI is a bubble. The only truly “cool” aspect of AI is how it integrates various existing technologies — like search, drawing, grammar correction, translation, and math to name some, into a single prompt. You type what you need, and it pulls from these tools to generate a response that wows those those who don’t know better.
Does it make life easier? Absolutely. Is it a revolution? Not quite. Will it save humanity? Hardly — unless you believe in masking and vaccination!
I don’t think it will make *life* easier except for those who are already wealthy. Those who need to earn income from labor will see like become more difficult: the question is how much more difficult?
I can’t find the tweet, but someone on X said “The purpose of AI is to give capital access to skills without giving the skilled access to capital.”
https://x.com/sonicminds/status/1835399293835899351 seems to be a retweet of a post by a deleted user @jeffowski. I think he fled to the safe space of Mastodon before the safe space of Bluesky became available. https://mastodon.world/@jeffowski says “BLACK for the death of American democracy” on the profile.
@Philip: I think the source I saw had a slightly different formulation. But regardless of the author’s/authors’ general political orientation, is it wrong?
Can AI do the job?
Does management think AI can do the job?
Does management think that AI will accelerate the job enough to allow for headcount reduction?
The answer, at least to #3, appears to be yes for now:
https://www.reuters.com/technology/ibm-pause-hiring-plans-replace-7800-jobs-with-ai-bloomberg-news-2023-05-01/
https://itmanager.substack.com/p/it-job-market-shrinking-will-that
“The job market which was robust when the pandemic ended has now gotten to the point where it continues to shrink. The cumulative loss in jobs since October of last year is 57,100 and 29,100 of that occurred in 2024.”
I have generally earned my living due to my ability to speak, write, and develop computer code. Now that those skills are worthless, I am worthless.
@FB, I used to blame computer processor annual development cycle driven by commercial considerations and George H Bush’s slow and Bill Clinton’s crony economy for need to become computer programmer vs scientist or R&D engineer. Later I learned about long term planning for “knowledge workers” which started during Nixon administration. I dreamed of ML that could do my job and better market for science and engineering and wanted to work on it but could not due to financial considerations. It is too late for me but it seems that science and engineering jobs with competitive compensation picked up since 90th which were absolute low, hopefully they will keep growing under new administration that will include Elon.
@perplexed
I studied engineering but would welcome a programming job!
The article that I cited shows an absolute decline in US IT jobs over the past couple of years coincident with numerous visas (the well-known H-1B and L-1 visas plus the UNCAPPED H-1Bs at nonprofits and the UNCAPPED THREE (3) YEARS of OPT which almost inevitably arises from each “student visa”).
I don’t see how that could be anything at all positive for an American worker.
FB: I think the quote is mostly true. I would take issue with the word “purpose” since the people developing LLMs don’t seem to be coordinated in any way or have any goal other than making money for themselves. It would make more sense to me as “the effect of AI…”
LLMs in coding couldn’t be happenning without free source community. Where else would LLMs be vacuuming code examples from to generate code boilerplates? So blame socialism.
Correction: In my post above, the statement “Does it make life easier? Absolutely.” should have read “Does it make life easier? Maybe.” I’m not sure how “Absolutely” slipped in — it completely contradicts everything else I said about AI in that post!
Regarding AI, H-1B visas, or outsourcing replacing local U.S. programmers, let me share a story with you.
About 20 years ago, at my Big company, we established a team in China, eventually growing to 50 developers, with the goal of transitioning development away from the U.S. Fast forward 10 years, and the entire outsourcing initiative was abandoned — today, there are no longer any developers in China. This is despite several of us making multiple trips to China, with some spending an entire month there to support and enable the new team. The primary issues? Maintaining consistent code quality and sticking to 8-hour workday, or even less.
About eight years ago, after losing hope in the China outsourcing effort, my Big company shifted focus to India, where we now have a team of around 30 developers. While the team in India is far more committed in terms of working hours, the challenge lies in the quality of the code. I frequently receive code review requests with major issues — no test coverage, no handling of side effects or edge cases, and minimal or poorly written comments. It often takes multiple rounds of reviews before the code is ready for approval. This causes significant overhead and missing deadlines.
H-1Bs? My Big company has not hired H-1B for over 20 years now.
Regarding AI, about a year ago, my Big company launched an initiative to implement Microsoft Copilot. Six months in, the project was discontinued without any explanation. Those of us who are seasoned developers, we saw issues with it and refused to use it. Meanwhile, less-experienced developers leaned heavily on Copilot, treating it like a copy-and-paste tool. We consistently rejected their code during reviews due to poor quality. When Copilot was deactivated, the seasoned developers were rejoiced.
About twenty years ago, I interviewed for an interesting contract programming position, with a defense/ security/control systems electronics company which used to be a household name but which name I can not recall now. After asking about projects prospects I was told that my job was to smooth out transition for new China-based programming team. Wondering how it was going to work out for national security I politely refused the job offer. Programming jobs market has had ups and downs in the 20 years passed but the company dropped off from public newsfeeds.
Why limit the work duties of a driving robot to just driving? It could also answer the phone (and deal with AI telemarketers), cook, go to work, and even do your taxes.
Perhaps the day when humans will have to masquerade as robots to survive is not that far away. Stanisław Lem, my very favorite science fiction writer, explored this theme in The Star Diaries.
Still did not get the jogger hands right. Also an extra tree shade, without a tree. That is the best we can expect from modern ML systems. Need to tone down AI hype and resurrect old expert systems concept, improved by applying statistical derivation methods developed by ML.
Is Thanksgiving post with turkey pictures and marinade/stuffing techniques coming?
You may know this but personal defence drones were are feature of Vernor Vinge’s https://en.wikipedia.org/wiki/Marooned_in_Realtime
Excellent series of books.
Talking about workers, I replaced my roof yesterday hear in Tampa because of the damage from Milton. There was no AI or robots involved. All the workers were non English speaking, just the supervisor knows enough English to communicate with customers. I asked him why can’t he/they use English speaking(US Citizens), he replied ” Let me know if you find any ” so i am still looking….
I call them rooftop Mexicans. Hard working devils.
A tenant paid his last month’s rent with a defunct Polaris 4-wheeler.
I put it on Craig’s and a couple of tough-looking young fellows showed up with cash money and slight accents.
“I’m guessing you fellas aren’t from around heah.”
“Kazahkstan”
They winched the wreck aboard their pickup and drove back to Billerica, Braintree, somewhere up there in terra incognita.
I had their e-mail and checked in to see how it had worked out. They’d fixed it and were blasting around some Mass. state forest.
Cossacks!
I’d say Caucasian with a touch of Asia around the eyes.
RH: We have a father-and-son roof repair team come and inspect our 21-year-old tile roof every year and fix any cracked tiles, etc. They’re both white native-born Floridians.
Note that if low-cost immigrant roofing labor weren’t available and that drove up the cost of roofing the result would be more people living in apartment buildings, which would be great for the environment, Climate Change, etc. The single-family house is something that will have to be eliminated if we are to continue to grow our population and simultaneously reduce CO2 emissions.
@PhilG All my insurance adjusters/inspectors were white native-born. Roof inspection and repair is a lot easier job than replacing entire roof in 1.5 days. I do not think Americans will move into apartments as a result of increased roof cost.
RH: Microeconomics 101 says that all economic decisions are made on the margin. Suppose that the NYC government decides to house one migrant at taxpayer expense in one NYC hotel room. If you say that this won’t have any effect on hotel rates for tourists and business travelers then it would also be true that taxpayers renting thousands of hotel rooms for migrants wouldn’t have any effect on hotel rates. If Maskachusetts can pay $3,000/month for an apartment for a single migrant family and that won’t have any effect on the prices of apartments for people who pay out of their wages then Maskachusetts can house every migrant family at $3,000/month and there will be no effect on the apartment market.
Consider a 2% increase in the costs associated with single-family houses that doesn’t apply to an apartment building and/or is greatly reduced in an apartment building (e.g., because 1,000 square feet of roof can keep rain out of 8 apartments in an 8-story building). If you say that this increase won’t motivate even one person to choose apartment living rather than single-family home living then you’re also saying that a 20% increase won’t motivate anyone to switch from shopping for a single-family home to shopping for an apartment in a mid-rise or high-rise.
(As the U.S. trends toward a long-term population of 600 million via low-skill immigration, my prediction is a dramatic increase in the percentage of Americans who live in apartment buildings. People whose average skill level is low can’t afford to spend as much on housing and all costs are lower in an apartment building (land, property tax, roof maintenance, insurance, energy, etc.).)
@PhilG I agree with you. But these theories rarely work in real life. Let’s say that all the construction is done by native born Americans. Then even the apartment prices will go up a lot. I’ve been following closely the consumer behavior for last 4 years. The economic theory says that once the prices starts going up, consumers should start buying less of these things or move to an alternative cheaper product. But in last 4 years consumers actually started complaining about the prices but their consumption of the same goods went up a lot. In theory, this should not happen. In my case I drink chamey blue reserve beer which I used buy for $22 for a 4 pack before the pandemic but after the inflation now it costs around $31 so I started buying a cheaper alternative which is three philosophers for 19 bucks. But by looking at the data consumers did not behave like this. I’m the only one odd man out here. So in last 4 years the housing prices went up so much but the average size of a house also increased. In theory, average size should go down so that you pay less.
RH: You’re certainly right that homo economicus from the Econ 101 textbook is not the same species as homo sapiens! If nothing else, it seems to take people a few years to realize that they can’t afford stuff anymore (having to declare personal bankruptcy being a wake-up call for many!).
I’m not sure that apartment prices would go up if construction were done by native Americans. If we assume that we need to import 30 migrants to get 1 construction worker, demand for apartments could raise prices more than the lowering effect of cheaper construction labor.
New houses are actually getting a little smaller right now: https://www.bankrate.com/real-estate/average-home-size/ “The median home size may have jumped by around 800 square feet since the 1960s, but in recent years, it has begun trending downward. Census data shows that, in 2021, new-construction single-family homes held to a median of 2,303 square feet. By 2023, that number had shrunk to 2,177 square feet.”
It may be that we’re setting market segmentation. New houses only for the rich and land becoming more expensive so it makes more sense to build a reasonably large house on an expensive-to-buy lot. We can see this phenomenon with Florida condos. The brand new ones sell quickly at breathtaking prices. The old ones are going down in value, partly due to expectations of expensive structural repairs required at the 40-year mark.
https://www.nytimes.com/2022/09/25/upshot/starter-home-prices.html : The disappearance of such affordable homes is central to the American housing crisis. The nation has a deepening shortage of housing. But, more specifically, there isn’t enough of this housing: small, no-frills homes that would give a family new to the country or a young couple with student debt a foothold to build equity.
The affordable end of the market has been squeezed from every side. Land costs have risen steeply in booming parts of the country. Construction materials and government fees have become more expensive. And communities nationwide are far more prescriptive today than decades ago about what housing should look like and how big it must be. Some ban vinyl siding. Others require two-car garages. Nearly all make it difficult to build the kind of home that could sell for $200,000 today.
“AI”, or at least LLMs are group think intelligence. They borrow the thinking of others and claim it as their own. Also, when they are wrong, they are not responsible. Thinking about it, who needs politicians!