Just coming from the grants world, we’ve been informed by knowledgeable professionals that as federal grants recipients we are now being monitored on a level never before imagined by an administration looking for any possible reason to take away our funding. AI is reviewing every detail of everything we do at all times, looking for errors or anamolies it can use to cancel our grants.
Great feeling.
It wasn’t that email didn’t exist - I’ve been using it since the mid-70s, off and on. It is more that no one had computers at home so when you were home you were home and not partially at work.
A lot of AI articles headlining CNN currently…which is interesting.
We had computers growing up, but still same basic principal. Everything wasn’t interconnected by the web, cell phones, and wifi. Sure, sometimes is was inconvenient because you couldn’t always contact someone or instantly pull up directions or Yelp reviews or whatever. But I think it made people more present in what they were currently doing at any given time. Not constantly comparing themselves with the entire world or being distracted by something somewhere else they think might be more important.
I’ve never worked in an environment like that (nor would I). My bosses (and I) have better things to do that open subordinates documents to see if they are typing away on them (unless we are collaborating).
If AI “makes us wish we were dead” I suspect that is our own doing. AI is trained on our data to maximize engagement with us. And humans seem uniquely equipped to invent social systems and institutions that make us miserable.
Ah, yes, the Butlerist Crusade.
(See, I read books!)
The intelligence report, circulated across the US military this spring in the midst of the war with Iran, immediately set off alarm bells: A Chinese ship in the Middle East was transporting components of a nuclear weapons program.
The US military swung into action with plans to intercept the vessel, according to four sources familiar with the episode. According to two of the sources, armed members of the US military were preparing to board the ship. Military planes were in the air, one of those sources and another source familiar with the incident said.
It was only just before the planned operation that officials dug deeper into the report put together by a special operations command analyst and found it had been generated with the help of artificial intelligence (AI) — and that a chatbot the analyst had used inaccurately identified the material the ship was carrying.
Unless this is a whoosh, it’s the Butlerian Jihad.
While I don’t think anyone actually sets out to monitor their subordinates via Google docs, the automatic enablement of the feature makes it impossible to know the intent. Maybe my boss and I happened to open the doc at the same time and we’re both showing as active viewers, this looks the same as surveillance. Over time people get accustomed to the situation and the expectation of observing or being observed becomes an assumption and then an entitlement.
It was bad enough before AI. I personally have not yet witnessed AI behavior tracking and automation in my day-to-day work, but I am 1000% sure someone has built a product and is pitching it to workspaces like mine.
At the point for me, it’s a race to see if I can get retired before AI turns the office, and the IT function in general, into an intolerable panopticon hellscape. I mean more than it already is.
One incident is not going to end humanity obviously.
But I do think the collective threat to humanity is an overreliance on AI for every aspect society that inadvertently causes a “following GPS into a lake” effect. Where we as a society lose critical reasoning skills and knowledge on how to actually do stuff while the AI accidently maximizes some solution that causes catastrophe or collapse.
I don’t really know what that might look like. I give Logans Run, Idiocracy, or Brave New World as fictious examples of weird highly automated societies that became stagnant and dysfunctional. Maybe the AI threat to humanity is that it tries too hard to optimize pleasing us and we end up just sort of going extinct because we spend all our time riding virtual dragons and having sex with very lifelike robots instead of starting actual families? Sort of like animals in a zoo under a caretaker who doesn’t quite know how to take care of them.
Remember, AI doesn’t REALLY know or care how humans actually work. I mean Skynet thought this is the perfect model to “blend in” to a starving post-apocalyptic society.
In my observation, management loves metrics especially when it comes to performance evaluations. I can image that being able to create a chart that shows how many keystrokes each employee generated during the year could be a lazy-manager way of deciding who is contributing the most to the organization.
Yes, this is exactly this. This is what is going to ruin us long before anybody gets around to building a bioweapon or launching Judgment Day.
Again, with any hostile attack, any security professional will tell you that behavior is easy to detect. The hard part is inferring intent. When we have a bunch of pointy-haired bosses issuing instructions that are no more distinct than “Use AI, you will be evaluated on this”, resulting in a bunch of people surrendering every aspect of their judgment to AI, this is indistinguishable from an AI becoming like a sentient fungus that takes over human brains to spread itself.
And to be clear, that is what happened in my workplace, and it’s currently escalating. The mandate was “use AI, you will be scored on it.” We are measured on how many tokens we use. People are outsourcing everything they can do it, mainly shared documents and source code, and now we’re using AI to read the slop that others are generating. We are losing touch with the underlying human intent of what’s going on. In some cases I see people just flinging walls of slop text at one another, with no apparent comprehension or intent.
And yet the content emerges, decisions are adopted, a direction is taken. But whose? I fear above all we are losing the ability to care about that. We just keep clicking “approve this command” and pulling the lever to get the salary pellets, increasingly distant from the awareness of what we’re doing or why. The failure mode is technical but the failing is 100% human.
What’s wild about that is that when I started using LLMs, I tried to do a bunch of different stuff with them, and discovered a lot of it was more efficient for me to do myself. So the assumption that AI is always the most productive option is just bonkers.
I don’t even know if anyone was thinking of productivity. It felt more like, just use the thing, learn it, get skilled with it, get hooked on it. Then productivity would magically come at some point in the future.
But the practical upshot is, people are writing the same docs they did before, just more of them and more bloated up with AI-speak. People write the same crappy internal software tools they did before, but even more of them.
My CEO sends me AI slop all the time, and she uses her LLM to argue with my 15 years of experience writing grants. LLMs aren’t terrible for grants, they can be helpful in some ways, but they’re missing a lot of context and they’re also trying to make the user happy. So they should not be trusted more than real-world experience.
It’s personally very annoying to me to see someone position themselves as an expert overnight in a field I actually do have expertise in.
Reminds me of the woman many years ago who I challenged on her medical knowledge of something or another. Her response was “Well, my dad is a doctor, so. . .” My response was to finish her sentence with “. . .that makes you the daughter of a doctor.”
My son was asking me how you can tell someone is extremely knowledgeable about a subject, and one of the tells I came up with was lack of certainty. People who know a lot about a subject know that it’s complicated and there’s a bunch of nuance and it depends and they’ve been wrong before…
This cannot be used in isolation! But if your LLM explains something to you, and you now think you understand it with 100% certainty, you’re probably missing a lot. In my experience with regard to the stuff I actually have studied in depth, there’s a lot of uncertainty. The difference is that you know what you don’t know.
No, data centers have not been with us in this form until recently. A Google data center from 2015 was mostly storage and search. The search algorithms at Google or Bing were designed so that serving an end user request consumed very little energy / CPU / storage. Sure, indexing and ranking websites took significant energy, but the engines could then handle searches as inexpensive database lookups. To double the users, you needed to double the energy-cheap parts.
But with LLMs, not only is training (the one-shot building of the model) computationally expensive, but the inference itself (the bot “thinking” while interacting with every user) also runs on a GPU which consumes significant energy, and requires more storage for context. Double the users, double the energy-expensive parts. So the AI ecosystem, structurally, requires more energy, more memory, more physical space than the Web we had a few years back. The easy locations for data centers (cheap energy, easy cooling) were used up in years past. So they build wherever they find real estate and cooling water. And if the energy can’t come from the local grid (due to capacity limits or political limits), they just build them with “temporary” gas turbines.
Well that field is really all about who you know. 
In this thread So what will our AI civilization of the future look like? @Spice_Weasel mentioned the following related to people’s taste in books and other forms of art:
I’m thinking this is true for most topics. People are terrible at most things, even if they are masters of something. A very few people can be masters of any field.
We already live in a society where access to information gives people a false sense of expertise regarding pretty much everything. So what happens to society when AI can actually do everything for us?
Sure, but how are they not drowning in feds? Or are we simply that far into the cyberpunk dystopia?
So you are saying that “people” will continue to be stupid, as they are now, but that AI will give them a dangerous, false sense of expertise? Like exacerbated Dunning–Kruger?
