What If The Hand Loom Weavers Were Children of the Nobility?
Gato Malo had a piece the other day that fit in with some of my recent thinking on AI and changes in the workforce. Most of hispost is about how the market values and pays for labor, and I mostly skimmed it because this way of thinking about value is as natural as breathing to me**.
But it was something he wrote near the end of the article that caught my attention (the lack of caps is gato's long-time style)
over the next 5 years, doctors, lawyers, engineers, accountants, and money managers are all going to come into crosshairs and see industries and payscales demolished. there will be a fearsome rear guard fight from guild systems (the bar, medical boards, CPA association etc) but it’s doomed to fail as they are just more unions trying to keep wages too high and prevent progress and increased access. taking humans out of medicine and breaking the cost spiral is the only possible way for the west to survive the already arriving wave of unfunded entitlement, and in the end, needs must will out and fiscal reality will trump guild control.
this is actually going to be amazing for consumers, a drop in price and expansion of availability in professional services that will rival the gains of the industrial revolution. once, making a pair of shoes consumed so many skilled resources as to cost months of wages. a good shirt was so valuable that people willed them to their children. now shirts are so cheap as to constitute less than an hour of most people’s labor. but the cost of “healthcare” has not worked like this despite being a technology product. law and accounting-services remain domains priced like rare gems.
they are about to be disrupted.
many are lamenting this forthcoming employment apocalypse for the over-educated, but we really should not. it will be amazing for the consumer and just like the tractor and the plow and the combine it will allow all manner of new and better innovation increasing standards of living through productivity gains as resources free up and move to profitable application.
A couple of weeks ago, for our anniversary, my wife took me to a very nice restaurant in a 5-star resort somewhere around Newport Beach in CA. At the table of 8 next to us were apparently several generations and branches of a single family who talked fairly loudly and included several people who mentioned Yale two or three or twenty times during the evening. It was a small random sample from the closest thing the US has to a governing nobility, the Ivy-League-educated elite. And much of the discussion through the evening seemed to focus on the family's fears of AI and its threats to their children's future job prospects.
I agree with Gato that AI has a huge potential to disrupt current work patterns, in the same way that the industrial revolution did. The 19th century disruptions were severe, and many people suffered as their experience and skill set no longer matched the new economy. But eventually everyone, from the poorest to the rich, were better off for letting the industrial revolution run its course.
But in the 19th century, the disrupted were essentially powerless. What happens this time around, though, when the disrupted are the ruling elite themselves? These potentially disrupted professions include lawyers and doctors who already have shown themselves very willing to organize to block innovation, squash competition, and protect their high pay. Just look at the history of the attempts by Congress to reduce Medicare reimbursements to doctors. And that was minor compared to the potential AI disruption. Let me give you another example of the powerful resisting a technological change that should have disrupted their businesses.
When TV first was being rolled out, the industry coalesced around a network of local broadcast stations, many of whom became affiliates of a network like NBC or CBS. Why this model? Mainly it was driven by technology -- the farthest a TV signal could reasonably be broadcast was about 50-75 miles. Thus everyone by necessity got their TV through three or four TV stations in their metropolitan area, each its own small business.
Now fast forward to today. There are multiple ways to broadcast a TV signal nationwide -- there are several satellite options and many streaming internet approaches. So now when we watch DirecTV or Youtube TV, we just watch the national NBC or ABC feed, right? Nope. Federal law requires that whatever service you use MUST serve up NBC, for example, via the local affiliate. That is why your streaming TV service harasses you when you travel, because it is worried about violating the law by showing you the Phoenix CBS affiliate when you are staying overnight in Atlanta (gasp).
This is hugely costly. In order to be able to provide NBC among its stations, Youtube TV must gather the feeds from 235 different stations. In the Internet streaming era this is costly but in the satellite era it was insane. DirecTV, with its limited bandwidth, had to simultaneously broadcast 235 stations, most showing identical content, just to legally provide you with NBC. So why this crazy, expensive, insane effort? I am sure you have guessed -- pound for pound local TV stations are among the most powerful lobbyists in the country. First, they have money and a massive incentive to defend their local geographic monopoly -- Car dealers and alcohol distributors are much the same, which is why every potential innovation is resisted in those markets. But TV stations have one extra card to play -- nearly every Congressman in the House likely depends on the three or four TV stations in one major metropolitan area for a huge part of their publicity and coverage. No politician is going to screw with that. At the end of the day, local stations did not get disrupted, they actually became more valuable with this government-enforced distribution of their product.
This is a small example of the fight that is coming in AI. Congressmen will couch their arguments in fear-charged terminology as if their real fear is some Terminator-like AI apocalypse. But the real concern will be from the influential elite who are being disrupted. What would have happed to the Industrial Revolution if the hand-loom weavers were the children of the nobility? Would the government have allowed the revolution to proceed? We are about to find out.
** footnote: The only thing I would add to Gato's discussion of labor and value is that I think a lot of our conversations are hamstrung by the multiple meanings of "value" and "worth". Gato correctly makes the point that one's own self-worth and how one values his or her own labor has nothing to do with how the market pays for one's labor. But the reverse is true too -- how the market pays for your labor should not necessarily have any bearing on how one values oneself or one's labor. I saw a beautiful production by the ABT of the ballet Giselle last night. Most of the ballerinas in the show don't get paid as much as a good plumber, but their labor is worthy and beautiful and -- despite the low pay -- many professional ballerinas love their (often short) life as a dancer. Most professional athletes before perhaps 1960 were the same, doing the thing they loved and finding fulfillment in excellence even when the market did not see fit to pay them much for what they did.
I'd love to see them at least get it to work, first. I periodically try to use AI for work. It gives me mostly wrong answers, but will usually give me a link to what I'm looking for. Right now it's just a better search engine. It can make top performers in various fields more productive. It won't bring mid-tier performers into high-tier (or low-tier into mid-tier) because in order to make good use of it you have to be smart enough to know when it's wrong. This is a big problem, and it doesn't look like it can't be solved by burying it in money even though people are trying really, really hard to do this.
We will almost certainly need some fundamental breakthroughs at both the hardware and software levels before we can move past this. The startup Entropic AI has some very interesting stuff at the hardware level, but even if it pans out it's still many years away from replacing Nvidia and AMD. On the software side, lots of very optimistic promises but with very incremental improvements actually delivered.
The Economist had an interesting article recently about AI and ad-supported websites - traffic is dropping off quickly because people look at the AI results and don’t visit the site. Google will be OK because they can put ads in their AI responses. Sites like TripAdvisor and WebMD are doomed.
One central assumption to these arguments is that AI has value that just needs to be monetized. As someone who worked in the field for a few decades, it's dangerous to jump from "these models sound intelligent!" to "there's real tangible intelligence to be applied here".
The value to be unlocked from AI for the foreseeable future is in automating tasks involving language where precise and correct results don't matter much (illustrations based on text or rough translations are excellent examples).
Healthcare, education, law and engineering are very fundamentally not professions that are bound to benefit from imprecise and flat out wrong results being expounded upon with high confidence.
>> this is actually going to be amazing for consumers, a drop in price and
expansion of availability in professional services that will rival the
gains of the industrial revolution
This seems a little blinkered as a prediction of what would happen in law and medicine.
A different model might say that all of the same services will be produced, but at lower cost. That's the kind of result you'd expect if the amount of "total medical service" currently being produced is more limited by the global store of knowledge and technology than by the number of workers in the field. Your chronic pain will continue to go untreated because no one believes it can be treated. Lowering the pay of the person who tells you to suck it up doesn't change that. Increasing the number of people willing, or legally empowered, to tell you to suck it up doesn't change it either.
For the law, things are even less favorable to the "expansion of availability in professional services" view. Something I often think about is a scene from a Chinese economic drama called Like a Flowing River.
After a lot of wangling of meetings and conversations, our entrepreneurial protagonist gets into a meeting with a very successful developer about starting a joint project. He mentions some legal concerns.
The major developer responds that "if you want, I can draw you up a document of limited liability, but if the government takes an interest, the document won't mean anything in court".
It is my understanding that this broadly reflects the reality of the legal profession in China today: there are plenty of lawyers, but the job is not particularly prestigious, and the government feels free to ignore the lawyers if it wants to. Those last two points are probably related to each other.
If the job of being a lawyer in America is taken away from the ruling class, they can strip that job of the power it is currently believed to hold.
The public AI engines are there to generate publicity. There is no business model based on generating book reports and term papers based on a general data set.
Corporate clients pay for the collation of customized data sets (LLMs) and the AI engine is trained on these data sets. Add a set of custom prompts, language models, and logic rules, and you get a system that can handle tasks in a specific area of knowledge very, very well - with less "hallucination" and false answers.
When trained on x-rays and other medical imaging, these systems do very well indeed.
They also do very well when trained on tightly-structured legal documents and procedures.
At the very least, they will provide a more advanced starting point for the human practitioner, and support their decision making.
This is the real future for AI. These are the paying customers.
Regardeing the threat to the elites:
The most secure people are those with the most responsibility:
The prescribing physician
The architect and engineer who sign and submit construction plans
The CPA who certifies the tax return/annual report.
Etc.
Legal, social, business, and ethical pressures will require a real person in these positions... they will use AI, but even though the AI captures their decisions, they will not be replaced.
If the generational pattern is that hard working innovators give birth to Art History/Gender studies majors fit only for soft sinecures - nepotism will be severely limited. The emphasis will shift to meriticratic advance of those who can do the kernel of creative work that cannot be automated.
One likely social impact is the decline of feminism, which was linked to the industrial revolution and the post-industrial info/service economy.... but most 'empowered' women agree with Barbie that "math is hard" and are allergic to executive responsibility. Robotics and AI will eliminate the manufacturing and puffed-up clerical/admin positions where women predominate.
For some real turtle wrangling consider why physicians used to get paid so much, that's right, they had offices, office expenses, staff and of course, malpractice insurance. Then they all became corporate drones working at clinics. Things change but they still somehow, stay mostly the same. I've been waiting for the death of colleges and universities with the available online information vastly overwhelming anything educational one gets at a brick and mortar institution but we all know that is just the tiniest fraction of what people who go to college are paying for.