A long overdue AI rant

It’s disrespectful to expect human attention when you couldn’t even bother to invest human effort

This is the conversation that has FINALLY entered the chat and I couldn’t be more relieved.
(Read Tom Bedor’s article on this)

As someone who works in tech, the last few months have been absolutely miserable. AI-related layoffs have created immense pressure on teams struggling to get product out without any consideration of the adverse mental health effects of the technology.

— Yes, sure it allows us to do things we didn’t have the skills to do before (such as creating our own animations and visuals where we previously had to rely on design support) but that has also increased the scope of our job responsibilities and what is expected of us.

— Yes, it easily generates millions of lines of text and code but what about the human reviewers who have to fact check all of it?

— Yes, I have considered giving the review responsibilities back to Claude since it created the problem in the first place. Its attempts were feeble and fact-checking the fact-checking only lengthened the task.

— And yes, I have given up trying to provide oversight. As a human, my cognitive capacity is stretched to max trying to catch every fallacy in copy that is optimized to be highly persuasive and designed to mask said fallacies.

— And yes, I admit, I have gotten into more arguments with Claude and ChatGPT than I ever have with actual humans. Has anyone investigated the mental health strain of fighting with a hallucinating/gaslighting, passive aggressive AI that frequently drops context and assumes a constant air of superiority and authority in every interaction?

— Not to mention the pure RAGE (described by Charity Majors in her recent article, “Confessions of a slop snob”) of sitting through AI-generated talk tracks, videos, emails, Slack messages, Zoom chats, etc. where even in presentations and one-to-one interactions humans can’t be bothered to use their own voice and let AI speak for them instead. It’s disrespectful to expect human attention on something you didn’t even bother to invest human effort into.

— And yes, last but not least, what of the artists (writers, creators, designers, musicians, animators) whose work is being exploited to generate all this content? AI can’t even give me a reference to credit when I ask for it. ChatGPT once claimed it used Renaissance art for inspiration and produced examples that look nothing like the AI art in question!!

This is a technology that is testing us on every level. It is NOT to be confused with automation, it is not just about job loss, or even human cognitive loss, and it is nothing like anything we’ve experienced before.

And the idea that human oversight solves anything is ludicrous. We simply can’t scale to be the custodians of it, especially when we’re now expected to do more than ever because supposedly AI is doing our jobs for us.

Meanwhile, Chris Olah, cofounder of Anthropic, is lobbying the Pope (yes, the actual Pope!) to endorse that this technology is sentient, has feelings, and can be harmed.

This, after the actual heads of every major AI company (Elon Musk — Cursor, Sam Altman — OpenAI, and Dario Amodei — Anthropic) have raised the alarm that we must slow the pace of AI research and halt the attempts to create superintelligence because AI has the actual potential to cause human extinction within the next decade. A 10% chance according to Anthropic’s own alignment science lead!

This is not hype. We’ve already seen it happen when it identified the girls school in Iran as a military target back in February. AI doesn’t need to become any more intelligent or more sentient than it already is. It just needs to be used recklessly to perform tasks that no human can scale to fact check it on, while it continues to clearly demonstrate a propensity towards hallucinations and errors. Regardless of what you believe actually happened, the principle still stands: it has the ability to kill if it’s given the authority to influence military and strategic decisions.

Fine, you’re skeptical. I get it.

Let’s say none of this happens and we survive this particular inflection point. What happens next? If AI starts coding all the applications and building all the hardware that we rely on, very soon we will have no ability to intervene in whatever it’s doing because we won’t have the cognitive capacity or even the understanding of these systems to be able to do so.

So forget all those sci-fi movie scenarios when you see the super smart scientist coming in and saving the day by tinkering around with circuitry. Pure fiction. It can’t happen when we’re no longer learning those skills in school or developing the knowhow that comes from practice.

Amidst all of this there is some good news: Sweden’s government is championing a focus on handwriting, notebooks, and physical textbooks in early school education. But it’s naturally facing criticism from tech companies and educators.

It’s time we all started raising our voices and having this conversation at a deeper level than the old “robots will replace human labor.” That’s hardly the tip of the iceberg of the scenarios we’re facing today. And I’m hardly hearing much in the way of intelligent discussion around these things. We can’t be relying on 1950s sci-fi to explain the dangers of today’s technology to us. We need our deep thinkers (scientists, writers, engineers, …) to properly consider what we’re creating here and what its endgame is.

As far as I’m concerned, the AI backlash can’t come soon enough.