AI stole my job and my work, and the boss didn't know – or care
theregister.com
theregister.com
Most of the content being manufactured today is quite lazy and I'm fine with AI writing it because I won't see it anyway.
But replacing a person without helping them grow into a new and more impactful job isn't the way to do this. I wish more companies would 'fire' certain types of work and help their employees grow into other roles.
AI doesn't have to be good to destroy this income. This is making it harder for creativity to thrive. Relatively few budding artists can count on private patronage or government sponsorship, and both of these come with plenty of strings attached. And pie-in-the-sky ideas like universal income are not going to wish the problem away.
Every doomsday prediction is wrong… until the final one isn’t.
It follows that mass-AI adoption should tilt the creative economic model more towards state-funding, ideally supported by a tax on AI usage. (i.e. taxing and reallocating some of the value saved by using it, to redress the things it's probably destroying)
Curious to see how you'd write this legislation. Because I know a lot of programmers and lawyers, and they're all a lot better at finding loopholes than preventing them.
Ideally metered by perf-per-energy-per-time.
You'll tax out some low-value use cases, but that seems less concerning as long as the tax rates are kept modest (<10%).
That would simplify collection and traceability.
And a bit of extra pressure on decentralization wouldn't be the worst thing.
Why should the state be funding artists in the first place? If the public can't be convinced to patronize artists directly (eg. buying their stuff), what makes you think they'd want to be compelled to fund artists via taxation?
And because when you run a capitalist economy, things that aren't profitable... don't spontaneously spring into existence at scale.
Unlike the travesty we made of rolling out social media, maybe we should think about what AI will cannibalize in the future, what we'd like to keep, and set up systems to financially support those things?
Personally, I'd rather not live in a future where all music, art, and news is the same milquetoast pablum regurgitated by an LLM.
But that means finding an economic model where not-that is feasible.
Is it? What are we going to miss out on if art's purely commercial? The most visible art that the public sees is already commercial, whether it's music on spotify or movies in theaters. Hell, even most art that's in museums are commissions funded by wealthy patrons. If anything, government funded art is a recent phenomena, and we have gotten along just fine without it.
Opinions that are commercially unpopular, and the larger and more diverse talent pool to generate art of all sorts.
Patronage is a corrupting influence, and I'd feel much better about most art being funded through public grants (even via a baroque allocation system) than by Starbucks.
Because it's funny, labor organization never seems to come up as a theme in Starbucks-sponsored art...
> If anything, government funded art is a recent phenomena
That's stretching the truth a bit, or are we not considering the Catholic Church or Medicis government? Or not recent?
(to pull the most pop-culture examples, without delving into the actual royal courts/republics)
There are plenty of historical examples of art that opposed the government and likely played a part in changing public opinion. There's a pretty reasonable argument to be made that art plays an important role as a check on authority when needed.
Positions in it, all the way to the top, are regular jobs that get hired for, not political appointments. Write something into its charter about the funding never being conditional on content, only on some proof of ability. Make the amount of funding (for the organization as a whole) indexed to something meaningful—like inflation, or Congressional salaries, or something like that, with a multiplier based on national population—rather than having to be appropriated annually or at some fixed level.
It's not perfect, and it can't be, because Congress can always decide they don't like what it's doing and go back and meddle. But there are many safeguards that can be put into place to reduce the amount of direct government interference with such an organization—and those above are just what I could come up with off the top of my head in a couple of minutes. Given a committee with an actual mandate to assemble something like this, and people who are genuinely interested in doing so, I'm sure a much more solid framework could be ironed out.
But I do think it fills in a public need that the free market is blind to. And perhaps structurally incapable of addressing. (Bezos running a charity newspaper notwithstanding)
The NIH / DOE / NSF grant process has its trend-following and insider problems, but generally parcels out money in an acceptable way. It's not impossible.
(Edit: this submission seems timely and relevant - The Blue Collar Jobs of Philip Glass - https://news.ycombinator.com/item?id=41240152 )
Previous rounds of innovation and steep changes in work/jobs, all the way back to the industrial revolution, were both a) not seamless transitions and left huge swathes of highly skilled people impoverished for the remainder of their lives, and b) for those who did transition to new jobs, there were new jobs for them to transition to. Not everyone can learn to be an AI developer. What jobs does AI create? Is everyone who isn't well versed enough in IT infrastructure going to have get a job as a janitor working the data centers for Azure cloud? Those are going to be some clean bloody floors.
I get so tired of the rhetorical "this has happened before," no it really fucking hasn't. Not on this scale, and not when having money was so utterly critical to one's ability to survive.
Yeah but talk about terrible jobs.
If I were an artist I think I'd prefer doing commodity art pieces to pay the bills vs. working in a restaurant.
And yes, it's subjective. I know some people probably like waiting tables.
> > waiting tables
> Yeah but talk about terrible jobs.
From sibling comment to yours, https://news.ycombinator.com/item?id=41276998 :
> ...a great deal of implicit classism ('i don't want to do a job like that, therefore such jobs are worthless').
Easy to say until your profession is next up on the chopping block.
Concentrated power multiplies the force of its actions, whether (subjectively) good or (subjectively) bad. See the danger?
Maybe someday, someone in power will decide a job you gain fulfillment from is a terrible job.
Them market. If your job provides value, then you shouldn't have problem finding people to pay you to do it. If you can't... maybe we shouldn't keep it around.
>Maybe someday, someone in power will decide a job you gain fulfillment from is a terrible job.
What are you trying to get at here, that because losing your job sucks and anyone's job could be at risk, we shouldn't be eliminating jobs at all?
Well, if a job can be done by a machine, it should be done by a machine - after all, if there is one thing common across humanity, it is the desire for progress.
The problem is that unlike what was believed a century ago, automation did not lead to less work for humans at all, quite the contrary: our ancestors fought (often literally to death) for the 40 hour week, and that was enough to feed a family on. Nowadays both people have to work (way more than) 40 hours a week to afford a shack, and politicians wonder our generation doesn't have kids. Guess fucking what, why they don't. And ask yourself where the fuck all that extra wealth from all the women working full-time ended up, hint: not in your bank account.
Is that really progress though? Honestly, more and more jobs can be done by machines. But what about the fact that human beings need to connect and feel at least that they are contributing somewhat to society?
Progress is not synonymous with automation. And it's not universal across humanity. The Amish recognize that not all automation is good and they hold meetings to decide which technologies should be introduced.
Many people realize that there is an ideal point of automation -- and complete automation is not ideal.
Well, if people's basic needs would be taken care of (=UBI financed by taxes on automation, or by just plainly taxing the rich), they could have time left for actually contributing to society - anything from caring for the elderly to making art.
That is an assumption that is not obviously true at all. Plenty of people are not creatives and not self-directed enough to do so. But even if they are motivated, can we really be sure that such a soceity would work, because people will know they are not truly needed?
To be honest, already, society seems rather meaningless in some ways and I think a lot of people find meaning in their work that they would not if no one really valued them economically.
And also, people may make art, but automation now also seems to imply that the world will drown in a sea of AI-generated art that may make actually getting such art to a wide audience almost impossible.
To me, it is far from clear than a totally automated society will be any sort of utopia.
Some people, when given the resources to thrive without having to work, take a bunch of time off in order to heal from long-term burnout. But even most of them will eventually return to doing something productive of some sort.
But does "society" have to happen only "at work"?
I agree that more stuff should be automated, but you’re so very wrong on this one that you should spend more time with human beings to understand why you’re so wrong. Ironical for someone who’s advocating for automation.
This is an opinion, although I suspect you hold it strongly.
> after all, if there is one thing common across humanity, it is the desire for progress.
How do you define progress? Pure production and consumption? Net increase in subjective happiness across all people?
If I can make a book recommendation, check out the Culture series by Iain M. Banks. These are well-trodden thought experiments in science fiction, and they're more relevant than ever.
Sure, but what is "progress"? Who gets to define it.
Are we sure "if a job can be done by a machine, it should be done by a machine" is universally true -- and even if it is, who and what says that that's necessarily equal to "progress"?
However, the transition took many many years to happen. Take an example a conductor in a bus to collect the fare, took many years to replace and via stages.
With AI, there are so many jobs (I agree shouldn’t exist) that had been impacted really quick before even preparing to learn a new skill.
Another problem is you don’t expect everyone to learn at the same pace.
Those answers were previously (occasionally)contained within succinct articles written by someone who did some research or was an SME.
Maybe it was the case that content farms were degrading the ability for information to be found on the internet, but AI has basically dropped a nuke on search results.
It is a wasteland and nothing is alive there.
The second wave will involve 3d modelers, animators, post-processing jobs in cinema.
I'm already seeing lot of developer jobs especially front-end, testing roles being replaced entirely. Ironically, only back-end, devops, cybersecurity appears to be safe as it is mission critical even when good chunk of the jobs are automated.
what leads to my conclusion is that your comment will hold true for the most part, much as people stopped buying newspapers once Google News took off, there will still be those that pay digital subscriptions for media outlets they support, in fact their market shares might increase as a result of creative destruction.
There are riches to be made during these chaotic times. There are none for those seeking to be a miner-for-hire anymore in a large chunk of the white collar space.
Can you give me some public examples if it's so prevalent? I play around with the AI Dev tools and I can't see them being a feasible replacement for a human yet.
I also have friends that are copywriters, illustrators, photographers and they're pretty much the same as they always have been with another tool under their belt. So if you could also substantiate this claim I'd be curious.
but if you think these new generation of AI tools are some innocent augmentative processes you are gravely mistaken. What you describe are the previous generation of AI tools which were using you as training. What you will see come out next year will be a truly reckoning force.
Not even Indians are safe.
Surely the company you worked for can't be the only one replacing people with these tools. What about the companies that actually have thousands of devs on their payroll?
> What you describe are the previous generation of AI tools which were using you as training
Do these new generation AI tools have names, mentions, parent companies? Or is everything in this shadow realm covered by NDA and I, as a mere mortal, cannot access?
> Not even Indians are safe.
What does this even mean?
> What does this even mean?
You know, I'm genuinely curious; do you honestly not understand what OP meant or are you actually disagreeing with OP by asking a rhetorical question?
Deisteve appears to be claiming that LLM-based techniques for automating programming will be cheap and effective enough that even that kind of outsourcing will be unprofitable by comparison.
It's like Tesla creating a full self driving car (level 5) and ending the promotional video with "Not even rickshaw drivers are safe". Just fluff.
> Neither are east european ones, Indonesian ones, or American ones for that matter. For the (unsubstantiated) claim that actual humans have been replaced.
The representation of Indonesian workers here is nil. Eastern European migrants are also unable to compete with the suppressed wages that Indian migrants can fill. It's an overwhelming motif playing out across all industries.
You can take a look at the scale of the exploitation of Indian migrant labour for reference:
This discussion is also away from my initial intent which was to find out any sort of source, whether it be a press release, a research paper, anything related to your claim of frontenders and testers being replaced.
I would still appreciate such an answer. Though I expect I'll be disappointed.
If something is simple enough to be semi-automated by writing scripts for low-trained humans, it's more likely to be simple enough to be fully automated by LLM "AI" than more complex jobs would be.
what if it was AI generated pornography?
but thats exactly why people are drawn to it
there is something about seeing shame, humiliation and violence in pornography that makes people tick
a big reason why most people dont really care about onlyfans
and why 90s to early 2000s pornography is sought after
But, maybe with AI testing becomes more valued. As who else will validate that whatever AI generated does what we actually want. And then reports the issues for fine tuning.
Doesn't make me feel much better though. If FE is on its way out I'd bet BE is no more than a year behind.
the end product, the UI, is limited in terms of edge cases
the backend however, there is no limit. business logic, automation, all of these are endless edge cases.
if any FE engineers thinking they should switch to BE, its too late. There is large generation of those that have been using PHP since early 2000s, have seen everything under the sun, and their experience is irreplaceable
Rest assured BE engineers, your golden era is coming. The era without FE engineers hounding you for API parameters and QA rejecting your tickets for some UI bug.
Same thing for devops and cybersecurity. Your golden era is coming.
We are close to having open source tech that is on par with closed source tech (flux vs OpenAI Sora).
I’m not referring the patent holding media giants. I’m referring creative studios.
It's been a few decades since I read New Scientist, for good reason. It was a fairly low quality publication - not unlike what Popular Science became.
If Cosmos is similarly poor, then this is not surprising.
https://web.archive.org/web/20220601000000*/cosmosmagazine.c...
There are two divisions of people in the AI war. People who want this because they think it’s the future so they are trying to apply it to everything. And people who don’t want or don’t care about LLM being added to their daily life.
LLM bots everywhere might not go away but neither will the push back against them. Hurt enough people with your attempts to displace people’s jobs and they will develop defenses as a reaction to the out right abuse of the internet’s good will.
So this proposal of closed internet and heavily walled gardens doesn't really work and it just leads to more sophisticated and harder to detect means of scraping that data.
Think about it, if you have data that is so valuable that people are willing to put a paywall and throttle their usage at the expense of user experience, its most certainly going to be dumped in the public.
Also this fear of LLM stealing your content is unwarranted. It's no different than somebody running compression algorithms on content they scraped so that its more accessible for the masses.
From my readings and speaking with lawyers it appears the current copyright laws won't stand with LLMs because of the difficulty of reproducing the entire "compression" process which requires extreme amount of money to replicate.
This is OpenAI and many other's legal defence. That there won't be attempts to replicate the underlying mechanism to prove copyright contents were used in the process.
A carefully written prompt to replicate copyright work like with Suno can claim compensation pro delicto but to prove every piece of content will be difficult to claim LLM or token based attention networks are able to "steal" without building it from scratch, training it on millions of hours, especially if its closed and patented.
So we might see people claiming victory over cases like Suno but they've already lost the bigger battle over LLM's legal status.
See I think that is only true for so long. People don't want easy ways to scrape data, they want to promote their efforts. Things could easily change which allow for easier identity verification. For example, Social Security numbers are seeing lots of criticism right now and there is call for a national digital ID that can be used instead. If a national ID were to be implemented it would offer a lane for something like verified access to happen. People already have a preference for platforms like Facebook where the farce is sold that it’s easier to identify real people by using it.
Then you tack that on with copyright infringement and blatant stealing that is happening with LLMs and there will be enough of a momentum for people to close off.
Especially on an internet where open platforms are constantly being attacked and devalued for being open. Content producers will only start sending minimal data to indexing bots, enough to get listed. SEO dependent sites will probably perish in this era.
Legal battles wont even matter if society shifts that way.
are you suggesting the same thing?
I am merely trying to illustrate how close that possibility is. The same way that you’re behind captcha wall for many sites. Imagine how that rots.
I realize it would be an arms race, but are there any browser plugins that can alert readers to content that is likely AI-generated?
I do think that some required disclosure that "this content was generated" might be a good thing. Sort of how magazine and newspaper advertisements that are formatted to look like news articles, or advertising programs on TV that look like news programs or talk shows, have to be disclosed as advertisements.
AFAIK bloom filter doesn't work for fuzzy matches, so any sort of light editing would defeat this.
his despite those tools being both unreliable and treating semantic correctness as equivalent to artificiality. In fact, the more skill and effort someone invest in writing clearly, organizing one's thoughts etc., the more likely their output is to be labeled as AI-generated; people are literally being penalized for getting too good. Increasingly this is being used in the commercial world as an excuse not to pay people for commissioned work.