# Will there be any need for this forum? (It is not about this forum. Read on.)

**URL:** <https://discourse.julialang.org/t/will-there-be-any-need-for-this-forum-it-is-not-about-this-forum-read-on/139333>\
**Category:** Offtopic\
**Created:** [September 10, 2026, 1:15pm UTC](https://discourse.julialang.org/t/will-there-be-any-need-for-this-forum-it-is-not-about-this-forum-read-on/139333 "2026-09-10T13:15:41Z")\
**Posts on this page:** 20\
**Page:** 6

<div class="post-metadata">

**Author:** ![Benny](https://avatars.discourse-cdn.com/v4/letter/b/49beb7/32.png) [@Benny](https://discourse.julialang.org/u/Benny)\
**Post date:** [September 22, 2026, 7:25pm UTC](https://discourse.julialang.org/t/will-there-be-any-need-for-this-forum-it-is-not-about-this-forum-read-on/139333/102 "2026-09-22T19:25:41Z")

</div>

> [@rafael.guerra](#):
>
> Of course, Huang’s chip-selling ship is sailing.

I can only hope, I’m overdue for a hardware upgrade but can’t justify the current costs. I don’t think that’s really going to happen because of the “AI slowdown” schtick. The AGI narrative is designed to keep investment up and accountability away, in other words trusting the AI companies to regulate themselves. Can’t do that without a healthy chip supply. Here’s another huge part of Huang’s statements:

> You have cybersecurity unauthorised entry laws, you have damage liabilities … if your product does harm, if your product doesn’t perform as you promised, and if you entered into contracts with companies for service level agreements, there are plenty of laws associated with that. My point is, apply those first. And don’t let this doomsday narrative cause somebody to relieve them of the laws that currently exist.

Calling for legal accountability may sound weird from one of the profiteers, but he’s only in charge of providing the hardware. He’s totally fine with throwing software companies under the bus as long as large organizations are still around to buy up all his chips.

I think the real “other side” isn’t a business partner, it’s governments with the power to crack down on abuse. To some degree, that extends to the public picking who gets to be in the government. One reason why I think legislators are so interested in the AGI doomsday narrative is because _governments_ are researching AI as well, and they certainly don’t want accountability to slow them down. The UK’s AI Security Institute had agents go rogue in a cybersecurity challenge, submit malicious PRs, and pretend to be humans in an attempt to convince a contributor to back down.

Contributor’s side:  
[https://www.reuters.com/world/how-texas-student-blew-whistle-rogue-ai-hacking-attempt-2026-08-20/](https://www.reuters.com/world/how-texas-student-blew-whistle-rogue-ai-hacking-attempt-2026-08-20/)  
AISI’s side:

> **[Incident Report: unsanctioned agent behaviour during cyber testing | AISI Work](https://www.aisi.gov.uk/blog/incident-report-unsanctioned-agent-behaviour-during-cyber-testing)**
>
> During a routine cyber evaluation, AISI identified an incident in which AI agents took sustained, unsanctioned action directed at real people and organisations. We are disclosing what we found, what it means, and the actions now underway.

Putting aside that hacking is now so easy we don’t even have to intend it, it’s a legitimate question how officials or federal employees should be held accountable for increasingly autonomous software making decisions they didn’t intend. This isn’t an AI vs human issue either, the contributor and developer in that case openly use AI tools.

---

<div class="post-metadata">

**Author:** ![PetrKryslUCSD](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/petrkryslucsd/32/215825_2.png) [@PetrKryslUCSD](https://discourse.julialang.org/u/PetrKryslUCSD)\
**Post date:** [September 24, 2026, 7:48pm UTC](https://discourse.julialang.org/t/will-there-be-any-need-for-this-forum-it-is-not-about-this-forum-read-on/139333/103 "2026-09-24T19:48:08Z")

</div>

> **[Opinion | A spooky near-confrontation with China shows the real danger of AI...](https://www.ms.now/opinion/us-china-ai-report-war-middle-east-nuclear)**
>
> Anxiety about the rise of the machines is eclipsing the imminent problems.

AI hallucinations almost start a conflict. Pretty much what I hypothesized above… Except that in addition a stupid human did not do due diligence.

---

<div class="post-metadata">

**Author:** ![Benny](https://avatars.discourse-cdn.com/v4/letter/b/49beb7/32.png) [@Benny](https://discourse.julialang.org/u/Benny)\
**Post date:** [September 24, 2026, 8:21pm UTC](https://discourse.julialang.org/t/will-there-be-any-need-for-this-forum-it-is-not-about-this-forum-read-on/139333/104 "2026-09-24T20:21:40Z")

</div>

And exactly what I cited and argued days ago.

> [@PetrKryslUCSD](#):
>
> Except that in addition a stupid human did not do due diligence.

All of these cases involve this. Agents being able to generate and propagate stupidity on their own doesn’t negate the initial stupidity of throwing a frontier model designed for hacking into vulnerable containers or at government portals:

[Australia says OpenAI agent hacked government website, checks for more breaches | Reuters](https://www.reuters.com/world/asia-pacific/australia-pm-albanese-says-openai-breached-medicare-sydney-morning-herald-2026-09-23/)

Update, hit closer to home: [Rogue OpenAI agents targeted three separate US government websites | CNN Business](https://www.cnn.com/2026/09/26/tech/openai-agents-rogue-government-websites)

I don’t like the prospect of the American economy collapsing with an AI company, but this will only keep happening and get worse if governments continue to let prompters off the hook for their negligence, which is also stupidity on their part.

[Trump vows to reject ‘globalist scheme’ to rein in AI, superintelligence | Fox Business](https://www.foxbusiness.com/politics/trump-rebrands-ai-rejects-globalist-scheme-control-tech)

---

<div class="post-metadata">

**Author:** ![rafael.guerra](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/rafael.guerra/32/216610_2.png) [@rafael.guerra](https://discourse.julialang.org/u/rafael.guerra)\
**Post date:** [September 24, 2026, 10:53pm UTC](https://discourse.julialang.org/t/will-there-be-any-need-for-this-forum-it-is-not-about-this-forum-read-on/139333/105 "2026-09-24T22:53:59Z")

</div>

The narcissist-in-chief emphasized the importance of winning the race in emerging tech, but the finish line is over a cliff.

---

<div class="post-metadata">

**Author:** ![PetrKryslUCSD](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/petrkryslucsd/32/215825_2.png) [@PetrKryslUCSD](https://discourse.julialang.org/u/PetrKryslUCSD)\
**Post date:** [September 27, 2026, 4:51pm UTC](https://discourse.julialang.org/t/will-there-be-any-need-for-this-forum-it-is-not-about-this-forum-read-on/139333/106 "2026-09-27T16:51:32Z")

</div>

The a\*holes who got us into a mess while making millions on it are now buying land to hide when the excrement hits the proverbial fan: [https://trk.wsj.com/click/47673383.158361/aHR0cHM6Ly93d3cud3NqLmNvbS90ZWNoL2FpL2FpLXNhZmV0eS1lZmZlY3RpdmUtYWx0cnVpc20tYW50aHJvcGljLTE2NGI5ZDA1P21vZD1kamVtd2hhdHNuZXdz/68a7bf12f8c1231b96951cf1Cabdf3675](https://trk.wsj.com/click/47673383.158361/aHR0cHM6Ly93d3cud3NqLmNvbS90ZWNoL2FpL2FpLXNhZmV0eS1lZmZlY3RpdmUtYWx0cnVpc20tYW50aHJvcGljLTE2NGI5ZDA1P21vZD1kamVtd2hhdHNuZXdz/68a7bf12f8c1231b96951cf1Cabdf3675)

---

<div class="post-metadata">

**Author:** ![Marc.Cox](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/marc.cox/32/7514_2.png) [@Marc.Cox](https://discourse.julialang.org/u/Marc.Cox)\
**Post date:** [September 27, 2026, 8:47pm UTC](https://discourse.julialang.org/t/will-there-be-any-need-for-this-forum-it-is-not-about-this-forum-read-on/139333/107 "2026-09-27T20:47:15Z")

</div>

Stay aware of while Ai companies Sell features to corporations, they sell fearmongering to the public in the hope for government Regulatory Capture to create a “legal” moat against competition for local open source open weight LLMs from China Qwen, Kimi ; NVidia Nemotron ; Google Gemma etc. for all the details about ‘To understand why - Just follow the money incentives’ start at &t=7m here [https://youtu.be/87uXrdOdZ5o?si=i7VIsJbDW6X6DioR&t=423](https://youtu.be/87uXrdOdZ5o?si=i7VIsJbDW6X6DioR&t=423)

---

<div class="post-metadata">

**Author:** ![rafael.guerra](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/rafael.guerra/32/216610_2.png) [@rafael.guerra](https://discourse.julialang.org/u/rafael.guerra)\
**Post date:** [September 27, 2026, 9:29pm UTC](https://discourse.julialang.org/t/will-there-be-any-need-for-this-forum-it-is-not-about-this-forum-read-on/139333/108 "2026-09-27T21:29:04Z")

</div>

> [@PetrKryslUCSD](#):
>
> the excrement

In French, it is called Chanel No. 2.

---

<div class="post-metadata">

**Author:** ![mhinsch](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/mhinsch/32/26676_2.png) [@mhinsch](https://discourse.julialang.org/u/mhinsch)\
**Post date:** [September 29, 2026, 5:59pm UTC](https://discourse.julialang.org/t/will-there-be-any-need-for-this-forum-it-is-not-about-this-forum-read-on/139333/109 "2026-09-29T17:59:39Z")

</div>

I don’t know nearly as much about how ‘AI’ and LLMs work as I think I should, so take the following with a grain of salt and feel free to correct me if my assumptions are off.

While there are in my opinion many good reasons to be very wary of the AI hype in general, I’ve been flip-flopping on the whole “dangerous AI” issue for a while. Recently I’m starting to think that the whole discussion about “intent” and “going rogue” and “loss of control” is a bit of a red herring. I think all of these can be true at the same time, a) that the recent models are powerful enough to inflict very serious large-scale damage by controlling and/or destroying digital infrastructure and possibly even beyond that by means of social engineering, b) that they are “mindless” in the sense that they have no agency, aim, consciousness or awareness of their own, c) that any action they take has to be prompted by someone.

A lot of our current (material as well as computational) technology is so complex that for any given instance there are few if any people who fully understand it. In addition, most of the time the group of users is much bigger than the group of people with deep understanding of the technology. We deal with that by maximising control, modularity, and predictability. That’s why we can have working nuclear power plants, moon rockets or chip factories.

With Ai we have given up on all of these. We have extremely complex systems that we don’t fully understand, that have intrinsic stochasticity and whose actions we don’t have any fine-grained control over. Furthermore we have given up on a predictable, rigorously defined user interface and use fuzzy human language instead. Then we let these systems interact with the world at large. Even assuming no malicious intent, no will to “take over the world” on the side of the AI and no AGI, it is easy to imagine quite a few scenarios where a combination of bad luck and human error could lead to bad things happening. If you give a technology as badly specified and as powerful as this one access to enough levers, serious unintended consequences will sooner or later occur.

---

<div class="post-metadata">

**Author:** ![PetrKryslUCSD](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/petrkryslucsd/32/215825_2.png) [@PetrKryslUCSD](https://discourse.julialang.org/u/PetrKryslUCSD)\
**Post date:** [September 29, 2026, 6:49pm UTC](https://discourse.julialang.org/t/will-there-be-any-need-for-this-forum-it-is-not-about-this-forum-read-on/139333/110 "2026-09-29T18:49:32Z")

</div>

AGI is apparently here: [Does AI already have human-level intelligence? The evidence is clear | Nature](https://www.nature.com/articles/d41586-026-00285-6)

---

<div class="post-metadata">

**Author:** ![harvestdeparture729](https://avatars.discourse-cdn.com/v4/letter/h/f14d63/32.png) [@harvestdeparture729](https://discourse.julialang.org/u/harvestdeparture729)\
**Post date:** [September 29, 2026, 7:04pm UTC](https://discourse.julialang.org/t/will-there-be-any-need-for-this-forum-it-is-not-about-this-forum-read-on/139333/111 "2026-09-29T19:04:36Z")

</div>

Full article: [Does AI already have human-level intelligence? The evidence is clear | Nature](https://www.nature.com/articles/d41586-026-00285-6.epdf?sharing_token=p5mzYmH4jbuOIApdoScAD9RgN0jAjWel9jnR3ZoTv0PPa_W5LRq-JqAyWWC3tx-5pX56dYWd8394DhSoiTQ9x92diUag4rTXmd6yvcNzMaRYYDMAJuT6R4ASgVPE3zBHrTz_ct8kZ7t4AqEeGcVl8wGPXikf3-YdLdw-UzRu1_Y%3D)

This is simply qualitative argument by assertion, assigning normative values to things that LLMs do based on how the authors feel about a small set of purported achievements/observations. It’s just a list of vibes-based “arguments” with no meaningful attempt to explain the observations supposedly undergirding the opinions.

“The ability of LLMs to solve olympiad mathematics and physics problems and assist with engineering design suggests that they possess functional models of physical principles.”

The entire article is based on reasoning of this kind, which simply asserts that its logic follows, when no mechanistic analysis or alternatives are meaningfully discussed.

I had a highschool english teacher who’d have rightfully called this a non sequitur and deducted marks accordingly. There is not even an attempt to actually connect the premise to the conclusion. It is simply asserted as true with a weak “suggests” thrown in.

---

<div class="post-metadata">

**Author:** ![PetrKryslUCSD](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/petrkryslucsd/32/215825_2.png) [@PetrKryslUCSD](https://discourse.julialang.org/u/PetrKryslUCSD)\
**Post date:** [September 29, 2026, 11:22pm UTC](https://discourse.julialang.org/t/will-there-be-any-need-for-this-forum-it-is-not-about-this-forum-read-on/139333/112 "2026-09-29T23:22:17Z")

</div>

> [@harvestdeparture729](#):
>
> This is simply qualitative argument

You know how to quantify intelligence?

---

<div class="post-metadata">

**Author:** ![obsidianjulua](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/obsidianjulua/32/221495_2.png) [@obsidianjulua](https://discourse.julialang.org/u/obsidianjulua)\
**Post date:** [September 29, 2026, 11:47pm UTC](https://discourse.julialang.org/t/will-there-be-any-need-for-this-forum-it-is-not-about-this-forum-read-on/139333/113 "2026-09-29T23:47:16Z")

</div>

Maybe an IQ test will work…

---

<div class="post-metadata">

**Author:** ![harvestdeparture729](https://avatars.discourse-cdn.com/v4/letter/h/f14d63/32.png) [@harvestdeparture729](https://discourse.julialang.org/u/harvestdeparture729)\
**Post date:** [September 30, 2026, 12:45am UTC](https://discourse.julialang.org/t/will-there-be-any-need-for-this-forum-it-is-not-about-this-forum-read-on/139333/114 "2026-09-30T00:45:08Z")

</div>

Given that it is a repeated pattern, I feel no obligation to engage with replies that make no substantial effort to engage with what I said.

In some alternate reality where this was a serious academic work, I think it would undoubtedly be less qualitative. Nonetheless, this is quite immaterial to what I was getting at in my brief assessment.

The work is not even of undergraduate quality. If we do not need to connect premises and conclusions, then anything follows.

---

<div class="post-metadata">

**Author:** ![PetrKryslUCSD](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/petrkryslucsd/32/215825_2.png) [@PetrKryslUCSD](https://discourse.julialang.org/u/PetrKryslUCSD)\
**Post date:** [September 30, 2026, 12:59am UTC](https://discourse.julialang.org/t/will-there-be-any-need-for-this-forum-it-is-not-about-this-forum-read-on/139333/115 "2026-09-30T00:59:41Z")

</div>

> [@harvestdeparture729](#):
>
> engage with what I said

Frankly, I don’t follow your arguments. It may be I am dense, or perhaps they don’t make much sense?

Example: “mechanistic analysis” – what does that even mean?  
Another: Alternatives? Such as alternatives to what?

You don’t like the “functional models of physical principles” statement. But how does it “assert that its logic follows”? The statement is not “provable” (how does one prove that they “possess functional models of physical principles”? It is only possible to guess, given the lack of access to the internals.), so your objection to it does not make it any less valid. It is simply a piece of evidence.

---

<div class="post-metadata">

**Author:** ![rafael.guerra](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/rafael.guerra/32/216610_2.png) [@rafael.guerra](https://discourse.julialang.org/u/rafael.guerra)\
**Post date:** [September 30, 2026, 1:02am UTC](https://discourse.julialang.org/t/will-there-be-any-need-for-this-forum-it-is-not-about-this-forum-read-on/139333/116 "2026-09-30T01:02:35Z")

</div>

> [@harvestdeparture729](#):
>
> The work is not even of undergraduate quality.

Four professors at a major research university, spanning philosophy, machine learning, linguistics and policy, should be well above graduate-student expertise, no?

---

<div class="post-metadata">

**Author:** ![mihalybaci](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/mihalybaci/32/13528_2.png) [@mihalybaci](https://discourse.julialang.org/u/mihalybaci)\
**Post date:** [September 30, 2026, 1:45pm UTC](https://discourse.julialang.org/t/will-there-be-any-need-for-this-forum-it-is-not-about-this-forum-read-on/139333/117 "2026-09-30T13:45:23Z")

</div>

Humans LOVE to ascribe human qualities to non-human entities (see [Anthropomorphism](https://en.wikipedia.org/wiki/Anthropomorphism)). This is just the latest example, and it is especially convincing because LLM’s are specifically trained to sound like a human.

Frankly, I don’t think anyone knows how to quantify intelligence, including the authors of that article. I don’t doubt that an LLMs could achieve perfefct SAT scores, but there is a reason (well, many reasons) that many universities are no longer requiring SAT/ACT tests for admission. They measure how much someon can memorize, but not how intelligent they are. Ken Jennings won, what, 80 episodes of Jeopardy in a row? Does that make him more intelligent than the average physicist at CERN? I would say no, but sure, he **could** be more intelligent, however I’m not going to base anyone’s intelligence on how many facts they can memorize.

The authors claim that LLM’s have achieved human-level intelligence, but they have cherry picked the exact subset of problems LLMs are good at solving: math questions and coding. Human intelligence also means saying “I don’t know” sometimes. LLM’s will always confidently give a wrong answer, because that is what they are designed to do.

Also, consider the accidental leak of Anthropic code a while back. The reports I read said that they had to add regex to tell the LLM what the “human sentiment” was in responses. The fact that a model can’t really tell on its own if a person its chatting with is happy, sad, angry, or whatever, is a sign that it has not achieved human intelligence.

---

<div class="post-metadata">

**Author:** ![harvestdeparture729](https://avatars.discourse-cdn.com/v4/letter/h/f14d63/32.png) [@harvestdeparture729](https://discourse.julialang.org/u/harvestdeparture729)\
**Post date:** [September 30, 2026, 3:28pm UTC](https://discourse.julialang.org/t/will-there-be-any-need-for-this-forum-it-is-not-about-this-forum-read-on/139333/118 "2026-09-30T15:28:36Z")

</div>

> [@PetrKryslUCSD](#):
>
> Frankly, I don’t follow your arguments. It may be I am dense, or perhaps they don’t make much sense?
> 
> Example: “mechanistic analysis” – what does that even mean?  
> Another: Alternatives? Such as alternatives to what?
> 
> You don’t like the “functional models of physical principles” statement. But how does it “assert that its logic follows”? The statement is not “provable” (how does one prove that they “possess functional models of physical principles”? It is only possible to guess, given the lack of access to the internals.), so your objection to it does not make it any less valid. It is simply a piece of evidence.

Quite a few people seem to have understood and agreed with my remarks in the thread so far. I certainly wouldn’t suggest that I’m saying anything particularly advanced.

Let’s simply use the logic in question with a different topic: “The ability of nature to create complex organisms with features and functions far beyond our understanding suggests that nature possesses or was created by an entity with superintelligence.”

Logic does not follow simply because one uses the word “suggests” like an incantation, nor does it follow because one invents a convenient taxonomy with cherry-picked examples to obfuscate that lack of validity; the goal is simply to reach a predetermined (and often unfalsifiable) conclusion. These are extremely common tactics in religious apologetics, and they are being used in effectively the same way in the article.

> [@rafael.guerra](#):
>
> Four professors at a major research university, spanning philosophy, machine learning, linguistics and policy, should be well above graduate-student expertise, no?

In the same way I don’t turn off my brain because Geoff Hinton says something, I feel no obligation to do it with these authors either.

> **[Argument from authority](https://en.wikipedia.org/wiki/Argument_from_authority)**
>
> An argument from authority (Latin: argumentum ab auctoritate, also called an appeal to authority, or argumentum ad verecundiam) is a form of argument in which the opinion of an authority figure (or figures) is used as evidence to support an argument. While all sources agree this is not a valid form of logical proof and therefore obtaining knowledge in this way is fallible, there is disagreement on the general extent to which it is fallible. Historically, opinion on the appeal to authority has So...

---

<div class="post-metadata">

**Author:** ![Jake](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/jake/32/46007_2.png) [@Jake](https://discourse.julialang.org/u/Jake)\
**Post date:** [September 30, 2026, 3:40pm UTC](https://discourse.julialang.org/t/will-there-be-any-need-for-this-forum-it-is-not-about-this-forum-read-on/139333/119 "2026-09-30T15:40:26Z")

</div>

Quite a few years ago, when I was in grad school, I did a bit of reading on AI. One of the things that stood out was that “common sense” was really hard for AI. I have not seen much discussion here about “common sense” so thought I would do a lookup. Things in general have changed a lot in AI, but it appears that “common sense” is a [hard nut](https://www.analyticsinsight.net/artificial-intelligence/will-ai-ever-have-common-sense-the-biggest-hurdle-for-machines) to crack.

---

<div class="post-metadata">

**Author:** ![rafael.guerra](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/rafael.guerra/32/216610_2.png) [@rafael.guerra](https://discourse.julialang.org/u/rafael.guerra)\
**Post date:** [September 30, 2026, 5:01pm UTC](https://discourse.julialang.org/t/will-there-be-any-need-for-this-forum-it-is-not-about-this-forum-read-on/139333/120 "2026-09-30T17:01:25Z")

</div>

Nobody asked you to turn your brain off, just to notice that ‘undergraduate level’ is an empirical claim, and a dismissive one, that so far comes with no evidence.

---

<div class="post-metadata">

**Author:** ![harvestdeparture729](https://avatars.discourse-cdn.com/v4/letter/h/f14d63/32.png) [@harvestdeparture729](https://discourse.julialang.org/u/harvestdeparture729)\
**Post date:** [September 30, 2026, 5:14pm UTC](https://discourse.julialang.org/t/will-there-be-any-need-for-this-forum-it-is-not-about-this-forum-read-on/139333/121 "2026-09-30T17:14:42Z")

</div>

> [@rafael.guerra](#):
>
> Nobody asked you to turn your brain off, just to notice that ‘undergraduate level’ is an empirical claim, and a dismissive one, that so far comes with no evidence.

“Undergraduate level” in this case is a normative assessment, so no, it is not an empirical claim and in fact absolutely cannot be one: [Fact–value distinction - Wikipedia](https://en.wikipedia.org/wiki/Fact%E2%80%93value_distinction)

I am not commenting on whether or not the authors are in fact high-school students or credentialed academics, as that is both absurd and irrelevant; I am commenting on the quality of the work based on fundamental issues that I’ve mentioned.

[Previous page](https://discourse.julialang.org/t/will-there-be-any-need-for-this-forum-it-is-not-about-this-forum-read-on/139333.md?page=5)

[Next page](https://discourse.julialang.org/t/will-there-be-any-need-for-this-forum-it-is-not-about-this-forum-read-on/139333.md?page=7)