Straight Dope 3/3/23: Followup - What are the chances artificial intelligence will destroy humanity?

Absolutely, I do too. My concern is about the rate of change vs. the rate of human adaption. We’ve seen that humans are very adaptable - to climate changes, migrating food, printing press, industrial revolution, but at are we up to the pace of AI advancements hitting critical mass?

The Internet has had widespread adoption for 30 years, social media for half that time, and we’re still struggling to deal with the polarization it brings.

What happens when a chunk of white collar jobs become obsolete in a single generation? That’s a lot of job re-training all at once.

As you point out you do observe boilerplate. In a job situation the products and routine are repetitive so the bots may converge on boilerplate. Perhaps not. It was just a thought.

The never-ending fear of AI or automation or machinery ‘taking all the jobs’ is the result of job losses being visible and somewhat predictable, but new jobs created from the opening up of the ‘adjacent possible’ that new tech brings are not yet known, because we have to discover them once the tech is here. But we always do.

What do these jobs have in common?

AI Engineer
Driverless Car Engineer
Cloud Architect
Mobile App designer
Podcaster
Podcast producer
Social Media Manager
Influencer
Telemedicine Doctor
SEO Specialist
Drone operator
Content Moderator
Content classifier
3D Printing Service Jobs
Sustainability manager
Diversity and Inclusion specialist
Uber driver
Digital Marketing Specialist
Big Data Analyst

The answer is that none of those jobs existed 20 years ago. Those, and thousands more.

AI is undoubtedly going to be disruptive. Lots of people will lose their jobs. But ultimately, if we can run businesses without huge numbers of office drones, labor and capital will be available to be invested in other things. That’s how economies grow.

In particular, AI has the potential to bring to the middle and working classes the kind of legal, tax, business, and other help that the rich use to magnify their power. AIs might be crap writers, but they will be great copy editors, researchers, cover artists and in other ways help disintermediate the real world like the web disintermediated publishing.

Starting a small business will require a lot less knowledge of how to navigate the labyrinth of finance, accounting, taxes, regulations, business rules, etc. That will make owning a business a lot more accessible to a lot of people. You’ll need a lot less capital as well.

And once the ‘truth’ problem is fixed, everyone will have access to a private tutor. Again, that mostly empowers the poor and middle classes.

I’d say it’s more a case of computers are doing things by raw number crunching that many people thought required general intelligence. If you go back far enough people thought doing differential calculus required insight into the nature of the problem to be solved, and were amazed that computers could be programmed with some rules of thumb that allowed them to converge on a solution.

I admire your positive zeal. But, I believe there is another side. Not just the artifacts of growth that we have seen before, but a more ominous social path that Robinson Jeffers captured in his poem “The Purse-Seine”:

“We have geared the machines and locked all together
into inter-dependence; we have built the great cities; now
There is no escape. We have gathered vast populations incapable
of free survival, insulated
From the strong earth, each person in himself helpless, on all
dependent. The circle is closed, and the net
Is being hauled in.”

How many things are there that we no longer have to do, but as a result, we no longer can do?

They (nearly) all look like travellers on the Golgafrinchan ‘B’ Ark to me.

The annoying thing for the Elon Musks of today is that things like Moderators of web sites will become more important, not less. (And more of those jobs will be salaried IMHO)

This is because while AI are becoming better at moderation, the current situation is to see the trolling humans trying (and succeeding) to get around the guardrails of the AI systems. As another Musk and Tweeter example showed. Human rights reps can not be dismissed with no repercussions, it shows that those jobs are also a “middle man” job that in reality is not going to be as easy to dismiss in the future as AI will need constant human input.

There will be a lot of work for humans to keep the internet, and many other systems, benevolent.

And everyone thought they were useless until their home planet was infected from a dirty telephone and everyone died.

There is another common trope in sci fi stories where, assuming AI doesn’t kill everyone, it basically results in a society where humans are more or less unable to do anything for themselves. Not even slaves or prisoners, per se. But effectively well cared for pets free to live out a vapid and pointless hedonistic existence (examples include Westworld, Wall-E, The Matrix, Logans Run, Zardoz, a few others I can’t think of at the moment).

I am not saying that AI is bad and we should end it – I would be out of a job then. My point is that we should prepare for the very real impact it will have on humanity. It’s interesting to discuss a sentient paperclip AI powered by a dyson sphere, but that’s not the danger we face.

It isn’t about the magnitude of the change; it is about the rate of change. There is no practical limit to the rate that technology, including AI, can change, but there is a limit to the rate an individual can keep up with. I acknowledged that there will be better jobs unlocked by AI. This isn’t about buggy whip manufacturers switching their product and the factory workers learn some new tools. It is about populations of workers having to job train every 5 or 10 years because their work keeps becoming obsolete.

And not just jobs, but arts. Books and cartoons are easy. How much longer until we push a button to generate a feature-length movie customized to a set of dropdown preferences? Maybe it will unlock a new era of post-post-post modern art that can only be created by humans. What risks does limitless AI-generated art bring?

And communications. Much of society is already at loggerheads because of a very limited AI at one very popular company. What are the issues when all news and analysis is AI-generated with all of the unconscious biases that we already have? Not just news and not just politics – everything. The sports page, the Idiot’s Guide to Surfing, your investment’s earnings report.

AI can unlock a lot of opportunities for the under-privileged, but human nature doesn’t seem inclined to sharing power. I’d be surprised if the ‘haves’ give it away with strings unattached. We’ve discussed how just having a bank account is key to upward mobility. What happens when everyone needs an AI trading bot for the 401k just to keep up? Better bots will cost more money and nothing will have changed.

I think there is a lot of speculation needed to imagine and then prepare for AI. It’s not a long-term concern. We’ll need a balance of cynics and optimists to prepare for the range of possibilities. There may be a lot of short-comings with GPT, but it shows how fast AI-based change can come out of nowhere.

“The Machine Stops” for the Ur-example

Perhaps this is why prepping and practicing survival skills is so popular as a hobby. There are a surprising number of people who know how to make flint knives and start a fire by hand with no prepared tools.

That’s amazing that story was written in 1909. A world where everyone is locked in their rooms with nothing to do but interact with a sophisticated machine! Preposterous!

Let’s also not forget the film Idiocracy as well. Pretty much everything in that society was run by automation too, which would be the only way such a society could survive.

I ran across this.
Imagine a rogue Zamboni!

A Zamboni is like a Roomba, it only has to work in a well-defined area doing a specific task.

Yes. I agree with your point, in principle, but you are not accounting for one factor: the guys out of jobs are not the guys getting the new jobs. I have seen time and again jobs in industry vanishing; the 52 year old welder at the now defunct shipyard didn’t get a job programming welding robots. That job went to a spry 29 year old with education and experience in software management. The skills the welder had were not the skills needed.

My dad was bought out to an early retirement when he was 58, in 1990. It was deemed better to pay him to stay at home and hire someone younger who already knew about this Windows thing, rather than trying to teach dad.
Of course, he was happy. But not all who are out of jobs are as lucky. Living on unemployment benefits suck.

Disruptive technologies sometimes find a better way to do something, and sometimes find a cheaper way to do something. These are not quite the same, since disruption is not always a net positive change. It likely saves money to manufacture in countries with lower wages on the other side of the world despite increased transportation costs and environmental burden. But this has domestic effects in terms of unemployment and local communities and taxes uncollected which may be made clearer during a hypothetical pandemic or supply chain crisis or changes in foreign policy. Sometimes disruption displaces inefficient laws and sometimes it lobbies to create illogical ones.

Agreed, they want somebody new who will grow with the company. But, there are still a few areas in the transition that require experience. I scanned the speakers list for the upcoming Automate Conference and toward the end there are non-degreed guys with specialties. One that caught my eye was ‘Visual Robotics Illumination Specialist’ No matter how wonderful the system is you still need a guy who knows how to shine a light in a hole.

Again, none of this is new. You know who had a really hard time adjusting to new jobs? The tens of millions of farmers disolaced by mechanization. A close second: the hundreds of thousands of telephone operators who lost their jobs to automated telephone switching. The farmers were not only untrained for other things, but they lived on farms and had to move to find work. The telephone operators went back to school or took clerical jobs.

We have been talking about automation ‘taking all the jobs’ since the Dope started. And it was already a constant debate when I was in college in the 1980’s. Fourty years later, and unemployment is at an all-time low - and we are still talking about automation taking all the jobs. In the meantime, we are still creating millions of new jobs per year.

The problem is that this time automation is coming for the chattering classes who have the megaphones, so the wailing and gnashing of teeth and calls for government action will be loud.

Too late.

As it has been the case for a while, it is mostly the Right wing the ones that complained the most about algorithms censoring them, and now demanding more government intervention to “free them”. The evidence remains shaky but that is not important for many right-wing sources, working the refs (AKA the government) is done then to gain even more advantages in the real world over perceived “enemies”. Nice racket from those sources.

I wasn’t actually talking about politics at all. The ‘chattering classes’ have historically been considered to be educated people in the arts, media, education, etc. People who have loud voices and megaphones and disproportionate power over the narratives that drive debate. They can be Republican, Democrat, or reformed Druids. Doesn’t matter.