# We need a "little gear" for AI news panic

**URL:** <https://boards.straightdope.com/t/we-need-a-little-gear-for-ai-news-panic/1033199>\
**Category:** The BBQ Pit\
**Tags:** ai\
**Created:** [September 19, 2026, 2:53pm UTC](https://boards.straightdope.com/t/we-need-a-little-gear-for-ai-news-panic/1033199 "2026-09-19T14:53:37Z")\
**Posts on this page:** 15\
**Page:** 3

<div class="post-metadata">

**Author:** ![QuickSilver](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/quicksilver/32/7832_2.png) [@QuickSilver](https://boards.straightdope.com/u/QuickSilver)\
**Post date:** [September 21, 2026, 1:06pm UTC](https://boards.straightdope.com/t/we-need-a-little-gear-for-ai-news-panic/1033199/41 "2026-09-21T13:06:57Z")

</div>

You go on ahead. I’ll join you later.

---

<div class="post-metadata">

**Author:** ![Jasmine](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/jasmine/32/2964_2.png) [@Jasmine](https://boards.straightdope.com/u/Jasmine)\
**Post date:** [September 21, 2026, 1:07pm UTC](https://boards.straightdope.com/t/we-need-a-little-gear-for-ai-news-panic/1033199/42 "2026-09-21T13:07:46Z")

</div>

> [@QuickSilver](#):
>
> And for fuck-sake, stop saying ‘sandbox’ like you know what a sandbox is.

Of course I know what it is! In fact, I know how to use it. You need a pail, a shovel, and a lot of imagination. A few toys, too. 😀

---

<div class="post-metadata">

**Author:** ![LSLGuy](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/lslguy/32/5813_2.png) [@LSLGuy](https://boards.straightdope.com/u/LSLGuy)\
**Post date:** [September 21, 2026, 1:47pm UTC](https://boards.straightdope.com/t/we-need-a-little-gear-for-ai-news-panic/1033199/43 "2026-09-21T13:47:36Z")

</div>

Cat turds optional.  
  
  
  
  
  
  
Actually, IME as a kid cat turds were mandatory. No way to keep the damned things out.

---

<div class="post-metadata">

**Author:** ![DPRK](https://avatars.discourse-cdn.com/v4/letter/d/4491bb/32.png) [@DPRK](https://boards.straightdope.com/u/DPRK)\
**Post date:** [September 21, 2026, 1:57pm UTC](https://boards.straightdope.com/t/we-need-a-little-gear-for-ai-news-panic/1033199/44 "2026-09-21T13:57:45Z")

</div>

> [@Jasmine](#):
>
> You need a pail, a shovel, and a lot of imagination

The sifter, do not forget the sifter!

---

<div class="post-metadata">

**Author:** ![QuickSilver](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/quicksilver/32/7832_2.png) [@QuickSilver](https://boards.straightdope.com/u/QuickSilver)\
**Post date:** [September 21, 2026, 1:59pm UTC](https://boards.straightdope.com/t/we-need-a-little-gear-for-ai-news-panic/1033199/45 "2026-09-21T13:59:30Z")

</div>

Oh, FFS.

It’s very cute that you’re able to get whatever AI agent you’re using to access your Access database that you stored in your onedrive to share with your work friends using your own credentials.

It’s a different thing entirely to use a copilot studio agent (yes, ChatGPT 5.1 is not bleeding edge) to access an enterprise cloud hosted SQL database and associated unstructured document repository, with proper authentication (not your own sysadmin rights), through a model driven app interface, and make that available to authenticated app users.

I’m not claiming to be a leading AI expert. I just use the tools that are made available to me by the customer’s IT infrastructure. And my experience has been that OOTB LLMs are clever, but require extensive custom tool building to do anything much more clever than summarize a document, and sure, modify a glorified spreadsheet.

And since you were so put off by my earlier response, please consider the fact that your personal experience with X is not representative of all experiences with X. I have very few fucks left to give for this type of pedantry. If that makes me seem cranky or annoying, that’s just the risk I’m willing to accept.

---

<div class="post-metadata">

**Author:** ![RitterSport](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/rittersport/32/6326_2.png) [@RitterSport](https://boards.straightdope.com/u/RitterSport)\
**Post date:** [September 21, 2026, 2:07pm UTC](https://boards.straightdope.com/t/we-need-a-little-gear-for-ai-news-panic/1033199/46 "2026-09-21T14:07:33Z")

</div>

I’m not sure what the point of this thread is, other than to scold people who disagree with you on the risks of AI, without you allowing any discussion or argument.

I don’t think it’s pedantic to point out that LLMs can definitely see DB models when you categorically said they couldn’t. Sure, add a bunch more conditions and maybe you’re right (I don’t really know, but you’re probably wrong, based on everything I’ve seen from you).

It’s unusual enough for an industry to plead with government to regulate them, and we don’t know what OpenAI and Anthropic are seeing with their most recent incarnations, but I’ll believe them when they tell me that, holy fuck, this shit is scary.

I’ve never heard of a marketing strategy that says, regulate us or we’ll accidentally destroy everything. Hey, buy our product!

---

<div class="post-metadata">

**Author:** ![QuickSilver](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/quicksilver/32/7832_2.png) [@QuickSilver](https://boards.straightdope.com/u/QuickSilver)\
**Post date:** [September 21, 2026, 2:17pm UTC](https://boards.straightdope.com/t/we-need-a-little-gear-for-ai-news-panic/1033199/47 "2026-09-21T14:17:40Z")

</div>

The point of this thread is to point out that rather than scarring the bejesus out of poorly informed public by effectively promoting AI doom-trolling, responsible news outlets should try to understand and educate the public of the nuances and where the real risks lie when it comes to AI.

---

<div class="post-metadata">

**Author:** ![RitterSport](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/rittersport/32/6326_2.png) [@RitterSport](https://boards.straightdope.com/u/RitterSport)\
**Post date:** [September 21, 2026, 2:30pm UTC](https://boards.straightdope.com/t/we-need-a-little-gear-for-ai-news-panic/1033199/48 "2026-09-21T14:30:16Z")

</div>

What are those real risks?

I see them as:

1. Giving bad actors everywhere the ability to hack into lots of places, including water and electrical infrastructure
2. Giving bad actors with access to a lab the ability to create bio weapons
3. Using AI in conjunction with our weapons systems to target “enemies”, with too much reliance on AI decision making
4. Using AI to make more and more convincing deep fake videos, making it easier and easier for bad actors to fuck with elections
5. Using AI to generate a mountain of realistic-seeming photos and other troll-farm crap.

Anthropic delayed the release of a newer version because it found so many zero-day vulnerabilities with existing systems that they had to give it to the companies first to fix up their problems. Anthropic and OpenAI are the “good guys”, if you can call them that. x.AI doesn’t give a flying fuck and is happy to churn out child porn all day long, and is happy to turn over their tech to the military with no restrictions on how it’s used.

(ETA: That’s leaving out Chinese AI systems, which are right behind the ones in the US, and then Russian, Iranian AI that we know nothing about, but much of this is open source, so they can do their own development and obviously wouldn’t stop it from attacking their enemies infrastructure)

Those are the issues today. Longer term, you have to worry about the reduction of the value of labor in favor of capital, as knowledge workers increasing find themselves being replaced. I hope this tech revolution works the way it has in the past, with new jobs created as quickly as they’re destroyed, but there’s no guarantee of that.

What do you see as the real risks?

---

<div class="post-metadata">

**Author:** ![QuickSilver](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/quicksilver/32/7832_2.png) [@QuickSilver](https://boards.straightdope.com/u/QuickSilver)\
**Post date:** [September 21, 2026, 2:59pm UTC](https://boards.straightdope.com/t/we-need-a-little-gear-for-ai-news-panic/1033199/49 "2026-09-21T14:59:43Z")

</div>

All of the above.

But for people who don’t think about this as deeply, the news just serves to panic them about the AI on their phone, and the reports that AI has hacked this, that or the other, without the nuances.

---

<div class="post-metadata">

**Author:** ![RitterSport](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/rittersport/32/6326_2.png) [@RitterSport](https://boards.straightdope.com/u/RitterSport)\
**Post date:** [September 21, 2026, 3:07pm UTC](https://boards.straightdope.com/t/we-need-a-little-gear-for-ai-news-panic/1033199/50 "2026-09-21T15:07:22Z")

</div>

…which is true for every subject that the news reports on.

But, giving bad actors the ability to hack our infrastructure, fuck up our elections, and release bioweapons seems existential enough that we should really be worried about those capabilities and if there’s any way to rein them in.

And, that’s just with today’s capabilities. The people seeing the next-gen stuff have their hair on fire and are quitting jobs and leaving behind a fortune in equity.

Three years ago, LLMs couldn’t count the number of Rs in “strawberry”, and now their hacking capabilities are so good that Anthropic and OpenAI are giving industries previews. Who knows where things will be three years from now? Five years from now? Definitely not you – you don’t even know the current capabilities.

The time to worry about these future capabilities is not when they’re already out there, destroying infrastructure, running on Russian and Iranian servers. The time to worry is before any of that happens.

---

<div class="post-metadata">

**Author:** ![QuickSilver](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/quicksilver/32/7832_2.png) [@QuickSilver](https://boards.straightdope.com/u/QuickSilver)\
**Post date:** [September 21, 2026, 3:27pm UTC](https://boards.straightdope.com/t/we-need-a-little-gear-for-ai-news-panic/1033199/51 "2026-09-21T15:27:19Z")

</div>

Who do you imagine yourself arguing with right now?

Pretty sure it isn’t me.

Maybe most people are smarter than I give them credit. Maybe when they hear AI hacked HuggingFace, and half dozen other things in the past year, they infer every possible threat you itemzied.

But I suspect not.

---

<div class="post-metadata">

**Author:** ![Gatopescado](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/gatopescado/32/5711_2.png) [@Gatopescado](https://boards.straightdope.com/u/Gatopescado)\
**Post date:** [September 21, 2026, 9:00pm UTC](https://boards.straightdope.com/t/we-need-a-little-gear-for-ai-news-panic/1033199/52 "2026-09-21T21:00:07Z")

</div>

I just watched a 2-hour episode of Independent Lense on PBS about AI, and now I’m even more confused.

🥴

---

<div class="post-metadata">

**Author:** ![QuickSilver](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/quicksilver/32/7832_2.png) [@QuickSilver](https://boards.straightdope.com/u/QuickSilver)\
**Post date:** [September 21, 2026, 9:05pm UTC](https://boards.straightdope.com/t/we-need-a-little-gear-for-ai-news-panic/1033199/53 "2026-09-21T21:05:48Z")

</div>

It’s a series of neural network tubes.

---

<div class="post-metadata">

**Author:** ![BigT](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/bigt/32/12044_2.png) [@BigT](https://boards.straightdope.com/u/BigT)\
**Post date:** [September 22, 2026, 12:12am UTC](https://boards.straightdope.com/t/we-need-a-little-gear-for-ai-news-panic/1033199/54 "2026-09-22T00:12:58Z")

</div>

I saw nothing wrong with this thread. The issue is very much overhyped in the media, jumping all over every little thing instead of focusing on the actual real threats from AI.

The only thing that bugged me a bit is that I’d use “low gear” in the title, not “little gear.” No one I know talks about driving in the “little gear.” But I assumed that was an Aussism. (or Kiwism? I forget where the OP is from.)

---

<div class="post-metadata">

**Author:** ![LSLGuy](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/lslguy/32/5813_2.png) [@LSLGuy](https://boards.straightdope.com/u/LSLGuy)\
**Post date:** [September 22, 2026, 12:28am UTC](https://boards.straightdope.com/t/we-need-a-little-gear-for-ai-news-panic/1033199/55 "2026-09-22T00:28:05Z")

</div>

OP mentioned upthread that “little gear” was a trump malapropism he was using mockingly.

And we thought GWB was bad at English.

[Previous page](https://boards.straightdope.com/t/we-need-a-little-gear-for-ai-news-panic/1033199.md?page=2)
