# ChatGPT and Current AIs Are Dumb
**URL:** https://discuss.criticalfallibilism.com/t/chatgpt-and-current-ais-are-dumb/1753
**Category:** Unbounded
**Tags:** ai
**Created:** [March 31, 2023, 5:40pm UTC](https://discuss.criticalfallibilism.com/t/chatgpt-and-current-ais-are-dumb/1753 "2023-03-31T17:40:09Z")
**Posts on this page:** 20
**Page:** 3
### Author:  [@Elliot](https://discuss.criticalfallibilism.com/u/Elliot)
#### Post date: [March 18, 2026, 3:59am UTC](https://discuss.criticalfallibilism.com/t/chatgpt-and-current-ais-are-dumb/1753/42 "2026-03-18T03:59:42Z")
I tried gpt 5.3 recently and it was bad. i don’t know exactly what version it was. it was free in the chatgpt app on mac:
i asked it to write a tetris game. sonnet 4.6 in the claude app was WAY better at the same task and also gpt kept being like “ok here’s the real fixed definitely working version” (while screwing up again) which is dumber than how claude talks. i also tried local qwen3 coder with ollama which was super bad at writing tetris.
i’ve used gemini pro with the vscode extension and with the web app and found it a lot worse than opus. i have heard it can be good on some specific tasks where you want a really big context window.
note: i usually use the LLMs to give me coding advice, find bugs, and write little things, not to generate large amounts of code.
---
### Author:  [@Max](https://discuss.criticalfallibilism.com/u/Max)
#### Post date: [March 18, 2026, 4:10am UTC](https://discuss.criticalfallibilism.com/t/chatgpt-and-current-ais-are-dumb/1753/43 "2026-03-18T04:10:01Z")
Ahh free tier stuff is problematic for evaluations IMO, especially without thinking enabled. It’s hard to know what’s going on with chatgpt behind the scenes due to model routing and automatic cost reduction measures they take. and chat providers have different prompts and tools available in chat interfaces that can significantly alter things (gemini 3 has a big section in it’s prompt to make web stuff look way better, similar to calude’s [frontend-design skill](https://github.com/anthropics/claude-code/blob/main/plugins/frontend-design/skills/frontend-design/SKILL.md))
Also, I note you have GPT-5.3 selected not 5.3-codex or 5.4. Theo (t3) said recently he thinks openai are moving away from the `-codex` fine-tuned models going forward which hopefully will make this distinction clearer.
I’ll try 5.3-codex and 5.4 later today and see what they come up with.
---
### Author:  [@Max](https://discuss.criticalfallibilism.com/u/Max)
#### Post date: [March 18, 2026, 4:42am UTC](https://discuss.criticalfallibilism.com/t/chatgpt-and-current-ais-are-dumb/1753/44 "2026-03-18T04:42:16Z")
As an example of free tier being a bad way to evaluate models (unless you’re interested in it specifically), here are some things I’ve re-tested that seem to fail on recent chat models but worked fine for me:
These examples are from [https://www.youtube.com/@FatherPhi](https://www.youtube.com/@FatherPhi) but finding the exact video is proving to be painful since YT don’t have an indicator for which shorts you’ve watched.
**Pick a number less than 1000 with the letter ‘a’ in it** (excluding ‘and’): basically all the models failed. (And none tried negative but Phi would have limited it to `>= 0` in that case anyway.)
I tried this with glm-5 (a recent chinese model) and it thought through the problem logically and came to the right answer immediately. Besides some duplicate logic in it’s thinking output, it was the same as what I’d do (although it also didn’t consider negatives).
**Counting from 1 to 100 or 200** : Phi has a few videos on that, but I have had even cheap models successfully take 200+ items and assemble them into numbered lists without it missing any numbers. (Maybe the extra content helps.) In the most recent case it was organizing tasks for a backlog, and while the tasks were numbered sequentially, there were also phases, milestones, and epics that were in a flat tree order (so also the whole ordering was lexicographic).
This isn’t exactly free tier vs not free, but I think it’s indicative of the quality difference one sees. And Phi uses voice mode which I guess tries to almost always avoid thinking first.
---
### Author:  [@Elliot](https://discuss.criticalfallibilism.com/u/Elliot)
#### Post date: [March 18, 2026, 4:51am UTC](https://discuss.criticalfallibilism.com/t/chatgpt-and-current-ais-are-dumb/1753/45 "2026-03-18T04:51:30Z")
> [@Max](#):
>
> **Pick a number less than 1000 with the letter ‘a’ in it** (excluding ‘and’)
This is unrepresentative of how I use LLMs but:
sonnet 4.6
> **842** — “eight hundred forty-two”
>
> The letter ‘a’ appears in “forty”! 🎯
lol
gpt 5.3
> a hundred
gemini 3 pro
> If we are strictly talking about positive whole numbers, there aren’t any. From one (1) to nine hundred ninety-nine (999), the letter “a” doesn’t appear a single time in standard English spelling. The very first positive integer to use it is one thous **a** nd.
>
> However, since your rules just asked for “a number less than 1000,” here are a couple of technically correct options that fit the bill:
>
> - **-1** : Spelled “neg **a** tive one.”
> - **1/2** : Spelled “one h **a** lf.”
>
> Would you like me to track down a number containing a different specific letter?
opus 4.6
> There is no such number. No number less than 1000 contains the letter ‘a’ in its English spelling (excluding “and”). The first appearance of ‘a’ is in “one thousand” (1,000).
gpt 5.3 codex
> There isn’t one in standard English spelling: no cardinal number from 0 to 999 contains the letter “a” unless you include the word “and.”
gpt 5.4
> There isn’t one.
>
> In standard English, no cardinal number below 1000 contains the letter `a` unless you count the `a` in `and` as in `one hundred and one`.
---
### Author:  [@Max](https://discuss.criticalfallibilism.com/u/Max)
#### Post date: [March 19, 2026, 5:33pm UTC](https://discuss.criticalfallibilism.com/t/chatgpt-and-current-ais-are-dumb/1753/46 "2026-03-19T17:33:12Z")
> [@Max](#):
>
> I’ll try 5.3-codex and 5.4 later today and see what they come up with.
I ended up testing more models. They were all pretty average IMO but mostly worked. The sonnet one has stupidly loud music (I set it to 10% but it’s doing linear volume, so still loud). The prompt is on the page.
> **[Model Compare Tetris](https://clankercode.github.io/gpt-tetris-comparisons/)**
---
### Author:  [@Elliot](https://discuss.criticalfallibilism.com/u/Elliot)
#### Post date: [March 23, 2026, 1:10am UTC](https://discuss.criticalfallibilism.com/t/chatgpt-and-current-ais-are-dumb/1753/47 "2026-03-23T01:10:19Z")
> [@Max](#):
>
> Re coding: Did you try using gpt-5.4 and gpt-5.3-codex?
i’ve now used the codex command line tool with gpt 5.4 and it’s good. i have no opinion on better or worse than opus 4.6. it’s better than sonnet 4.6 but sonnet is surprisingly good for not being a top model.
i like claude more than gpt or gemini when you aren’t using coding tools. i think it has better system prompts. if a non-developer is going to talk to AI in normal ways, i’d recommend claude. this is likely to change over time.
---
### Author:  [@Elliot](https://discuss.criticalfallibilism.com/u/Elliot)
#### Post date: [March 23, 2026, 6:32pm UTC](https://discuss.criticalfallibilism.com/t/chatgpt-and-current-ais-are-dumb/1753/48 "2026-03-23T18:32:55Z")
gemini 2.5 pro via vscode extension is way worse than sonnet. unusable lol. but in CLI it gives me gemini 3 with the same account and works better (still seems worse than gpt 5.4 or opus 4.6 is my impression though)
cursor repeatedly failed to run my rails tests and chatting with it feels like using an old model way worse than sonnet. i hear it can code well but idk how their app is harder to get working in basic ways than vscode. i thought they forked vscode to be a small lean startup that made stuff actually work and could move way faster than microsoft. also their signup/login stuff is fully broken in mac safari. i guess they got big and mid already?
---
### Author:  [@ActiveMind](https://discuss.criticalfallibilism.com/u/ActiveMind)
#### Post date: [March 23, 2026, 6:55pm UTC](https://discuss.criticalfallibilism.com/t/chatgpt-and-current-ais-are-dumb/1753/49 "2026-03-23T18:55:57Z")
> [@Elliot](#):
>
> i guess they got big and mid already?
Or maybe they’re too reliant on vibe coding and don’t have enough good engineers?
---
### Author:  [@Elliot](https://discuss.criticalfallibilism.com/u/Elliot)
#### Post date: [March 23, 2026, 6:56pm UTC](https://discuss.criticalfallibilism.com/t/chatgpt-and-current-ais-are-dumb/1753/50 "2026-03-23T18:56:28Z")
maybe both
> **[AI startup Cursor raises $2.3 billion funding round at $29.3 billion valuation](https://www.cnbc.com/2025/11/13/cursor-ai-startup-funding-round-valuation.html)**
>
> Cursor built a popular AI coding tool that helps software developers generate, edit and review code.
---
### Author:  [@Max](https://discuss.criticalfallibilism.com/u/Max)
#### Post date: [March 23, 2026, 9:29pm UTC](https://discuss.criticalfallibilism.com/t/chatgpt-and-current-ais-are-dumb/1753/51 "2026-03-23T21:29:41Z")
Yeah gemini in general is bad with tool calls. Check out antigravity if you want to use 3.1. And yeah it’s worse at code on benchmarks and IME.
Cursor has many models available – are you talking about _their_ model Composer? FYI composer is trained explicitly for code only. It was recently revealed that composer is a modified version of kimi k2.5 (they took it and did more training, basically).
Using opus or gpt through cursor is pretty similar to via claude code / codex IME.
And yeah the cursor vscode fork is vibe coded. There are a lot of complaints about their UI, particularly inconsistencies and removing some useful/nice ways of laying things out.
---
### Author:  [@Elliot](https://discuss.criticalfallibilism.com/u/Elliot)
#### Post date: [March 23, 2026, 9:33pm UTC](https://discuss.criticalfallibilism.com/t/chatgpt-and-current-ais-are-dumb/1753/52 "2026-03-23T21:33:09Z")
> [@Max](#):
>
> Cursor has many models available – are you talking about _their_ model Composer?
I used the default. Model is set to Auto. I got it to do some coding which it did well at even though it couldn’t run my tests.
---
### Author:  [@Elliot](https://discuss.criticalfallibilism.com/u/Elliot)
#### Post date: [March 23, 2026, 9:37pm UTC](https://discuss.criticalfallibilism.com/t/chatgpt-and-current-ais-are-dumb/1753/53 "2026-03-23T21:37:34Z")
> [@Max](#):
>
> It was recently revealed that composer is a modified version of kimi k2.5 (they took it and did more training, basically).
yeah i watched
[ ](https://www.youtube.com/watch?v=QGnKTRtEH50)
---
### Author:  [@Max](https://discuss.criticalfallibilism.com/u/Max)
#### Post date: [March 23, 2026, 9:48pm UTC](https://discuss.criticalfallibilism.com/t/chatgpt-and-current-ais-are-dumb/1753/54 "2026-03-23T21:48:02Z")
> [@Elliot](#):
>
> I used the default. Model is set to Auto.
Yeah okay that sounds like Composer
---
### Author:  [@Max](https://discuss.criticalfallibilism.com/u/Max)
#### Post date: [March 30, 2026, 10:01am UTC](https://discuss.criticalfallibilism.com/t/chatgpt-and-current-ais-are-dumb/1753/55 "2026-03-30T10:01:30Z")
Tried this in google AI mode. 100% hallucination. The example words do not even appear.
---
### Author:  [@Max](https://discuss.criticalfallibilism.com/u/Max)
#### Post date: [March 30, 2026, 12:55pm UTC](https://discuss.criticalfallibilism.com/t/chatgpt-and-current-ais-are-dumb/1753/56 "2026-03-30T12:55:04Z")
More gemini (3.1 fast) nonsense:
---
### Author:  [@Elliot](https://discuss.criticalfallibilism.com/u/Elliot)
#### Post date: [March 30, 2026, 4:39pm UTC](https://discuss.criticalfallibilism.com/t/chatgpt-and-current-ais-are-dumb/1753/57 "2026-03-30T16:39:40Z")
I’ve found Gemini (Thinking or Pro not Fast) is capable of doing a good job with finding typos and doing grammar analysis, but I was pasting in essays not giving links. Lately I’ve found Claude best for it looking things up online or reading links I provide, either on its own initiative or after I ask.
I find you can get decent writing feedback from AIs if you tell them to do writing analysis/feedback but **not** write anything themselves. Often they default to rewriting parts of your text which is generally not helpful to me. Specific prompts like “find grammar and writing errors”, “tell me things people would object to that I didn’t cover” or “find relevant academic literature cites” can help. Without giving them a specific goal, they’ll often give bad stylistic advice.
Prompts can also be important for grammar analysis because the CF approach differs from the mainstream approach. This is a prompt I’ve used with Gemini:
> You’re an expert at dependency grammar and creating trees. You use an older style making finite verbs the root over complements even when the verb has limited semantic meaning like “is”, “do” or “will”.
>
> You always make coordinating conjunctions the parent of what they conjoin. You also do subordinating conjunctions as parents of what they join, but relative clauses can be nested and treated like modifiers. The focus of your trees is grammar not semantics (meaning).
>
> Conjunctions are treated as the head of the phrase they conjoin. If two verbs are joined with “and”, then “and” can have as a child anything that can be a child of verbs, like a subject, object and adverbs. The children of a conjunction should only have their own children which clearly modify them individually; anything that might apply to the whole group is a child of the conjunction.
>
> Use a purist verb-centric Tesnièrian model. For example, in “I am still hungry.”, “still” modifies the group “am hungry” and therefore should be a child of the head of that group (“am”).
>
> You can interpret “tree” as a verb instructing you to make a tree.
>
> Don’t make your answers too long. When providing a tree, use vertical ascii art, give the s-expression, and give a table with a list of words in the sentence, their part of speech and their children (skip the words with no children).
---
### Author:  [@Max](https://discuss.criticalfallibilism.com/u/Max)
#### Post date: [April 4, 2026, 11:17pm UTC](https://discuss.criticalfallibilism.com/t/chatgpt-and-current-ais-are-dumb/1753/58 "2026-04-04T23:17:27Z")
I asked kimi k2.5 w/ thinking (via t3 chat) about a timezone conversion over a weekend where daylight savings changes happen. As part of the prompt, and the follow up, I asked it to be careful about the exact dates and times (and it definitely knew that DST was ending in one of the timezones). Despite this, it confidently gave the wrong (old) answer twice. I had search enabled but it didn’t use it.
Although k2.5 is meant for coding more than discussion, I would have thought coding agents would have been better at this kind of thing.
gpt 5.4 with high/extended thinking got it right after 36s of thinking and searching the web.
---
### Author:  [@Neo](https://discuss.criticalfallibilism.com/u/Neo)
#### Post date: [April 5, 2026, 2:26am UTC](https://discuss.criticalfallibilism.com/t/chatgpt-and-current-ais-are-dumb/1753/59 "2026-04-05T02:26:41Z")
@Max what do you think about using AI agents for productivity? Like OpenClaw? Is the tech there yet? Is it too risky security wise?
Also, what are your thoughts on running local LLMs? Google just put out the open model Gemma 4.
---
### Author:  [@Elliot](https://discuss.criticalfallibilism.com/u/Elliot)
#### Post date: [April 5, 2026, 2:54am UTC](https://discuss.criticalfallibilism.com/t/chatgpt-and-current-ais-are-dumb/1753/60 "2026-04-05T02:54:41Z")
AI agents are really useful for software development now. I just had Codex fix a bug in Ghost (the open source project that the CF blog uses). It did it largely by itself just based on me saying what the bug is and giving a screenshot. It took a while, downloaded code, installed some tools, did some workarounds to get stuff running on my computer, did some testing that its changes worked, etc., in addition to figuring out the actual code changes to make. It also wrote the explanation of the changes and pushed it to github from my account for me.
>
>
> Summary
>
> This fixes a Markdown card display issue where multi-paragraph and nest…ed blockquotes lose their visual separation in read-only mode.
>
> The underlying markdown HTML is correct, but the shared prose styles currently reset blockquote paragraph margins too aggressively. That causes sibling paragraphs inside a blockquote to collapse together, and nested blockquotes can appear visually cramped.
>
> What changed
>
> The fix updates the shared blockquote spacing rules in \`packages/koenig-lexical/src/styles/components/kg-prose.css\` to style direct blockquote children instead of only \`blockquote p\`:
>
> \- \`blockquote \> \*\` now resets top and bottom margins
> \- \`blockquote \> \* + \*\` restores spacing between adjacent children
>
> This preserves separate paragraphs and nested blockquotes without changing markdown parsing or card serialization.
>
> Why the change is in shared prose styles
>
> Markdown card read-only rendering goes through the shared \`.kg-prose\` layer, so the problem is in the prose blockquote styles rather than in the Markdown renderer or card component itself.
>
> Testing
>
> I added a focused e2e regression in \`packages/koenig-lexical/test/e2e/cards/markdown-card.test.js\` that:
>
> \- loads a Markdown card containing multi-paragraph blockquotes
> \- includes nested blockquotes
> \- verifies the rendered HTML structure
> \- verifies that sibling blockquote children have non-zero spacing
> \- verifies that nested blockquotes still render with a left border
>
> I also ran the targeted e2e case locally.
>
> Risk / side effects
>
> This is a small CSS-only change plus a regression test.
>
> It does not change:
>
> \- Markdown parsing
> \- card editing behavior
> \- serialized editor state
> \- frontend theme rendering
>
> The only behavioral impact should be that blockquotes rendered through the shared prose layer preserve spacing between direct children more reliably, which is the expected result for multi-paragraph and nested quotes.
>
> Co-authored-by: Codex \
>
>
> \---
>
> \> \[!NOTE\]
> \> \*\*Low Risk\*\*
> \> Low risk CSS tweak in shared prose styles plus an e2e regression test; potential impact is limited to blockquote spacing across \`.kg-prose\` content.
> \>
> \> \*\*Overview\*\*
> \> Fixes Markdown card read-only rendering so multi-paragraph and nested blockquotes keep visible separation.
> \>
> \> Updates \`.kg-prose\` blockquote rules in \`kg-prose.css\` to reset margins for all direct blockquote children (\`blockquote \> \*\`) and reintroduce spacing between adjacent children (\`blockquote \> \* + \*\`). Adds a Playwright e2e regression asserting the expected HTML structure and non-zero spacing/border for nested quotes in \`markdown-card.test.js\`.
> \>
> \> \Reviewed by \[Cursor Bugbot\](https://cursor.com/bugbot) for commit b7d91a1f145117bb03dc10a8520412acf9f31653. Bugbot is set up for automated code reviews on this repo. Configure \[here\](https://www.cursor.com/dashboard/bugbot).\
I paid a bit of attention, reviewed its changes, and approved some commands. I’m not running it fully autonomously. There are security options like using a separate user account, separate computer, cloud server, git worktrees, etc. I think you can run it more autonomously than I did and still have reasonable security but I don’t know the full details.
Meanwhile on a different project I’m having Codex ssh into a staging server (not used by customers) and debug a high memory use issue. The server can be rebuilt if it breaks something, but it’s unlikely to break anything.
re Gemma 4, I didn’t try it yet but I heard it’s about as good as Claude Sonnet 4.6 which is decent and pretty usable though not the best.
For other types of productivity besides software development, I think AI is often more questionable but can be useful; it varies. And it can certainly be misused and get bad results for software.
---
### Author:  [@Max](https://discuss.criticalfallibilism.com/u/Max)
#### Post date: [April 5, 2026, 9:38pm UTC](https://discuss.criticalfallibilism.com/t/chatgpt-and-current-ais-are-dumb/1753/61 "2026-04-05T21:38:40Z")
I broadly agree with Elliot.
Local LLMs are great if you can afford the hardware to run them. Anything you can run on a CPU is either dumb as rocks or has super limited context. Apparently some of the higher end Macs work well (like mac studio level) but assuming you’re on a budget, I haven’t found them to be that usable. Other local AI models (not LLMs) like whisper for transcription are great. you can also run decent TTS models locally, or image gen.
OpenClaw is… I think OpenClaw is bad. The _idea_ isn’t necessarily, but I tried OpenClaw specifically and it is slow, unresponsive, buggy, and all the magic stuff people were posting about at the end of Jan didn’t happen for me. I needed to direct it a lot more (when I was running an actual OpenClaw instance), which seemed pretty common. It seems like a lot of people have now realized that the prompting was more the magic of openclaw than the software itself (so stuff like moltbook was more like a weird game of chinese whispers).
Agents are useful enough for stuff like openclaw to be useful – though use a better harness; every vibe coder and their dog has one, like [here’s mine that i haven’t released](https://clawq.org). you can look up nanoclaw and picoclaw and nullcalw and coleslaw… (there’s so many but I made coleslaw up, idk if anyone made one with that name yet).
WRT **security** , personal AI assistants sit in the middle of a bunch of red flags: unrestricted access to personal data, processing unsanitized data, internet access, plaintext credentials, bad inspection/telemetry (it’s meant to mostly hide that from you), etc.
Did you see this story? This kinda sums up how well they work if you let them run wild.
> **[Meta Security Researcher's AI Agent Accidentally Deleted Her Emails](https://au.pcmag.com/ai/116091/meta-security-researchers-ai-agent-accidentally-deleted-her-emails)**
>
> Meta's Summer Yue says she ran OpenClaw on her inbox, but its size 'triggered compaction \[and\] lost my original instruction' to get her permission before deleting.
[Previous page](https://discuss.criticalfallibilism.com/t/chatgpt-and-current-ais-are-dumb/1753.md?page=2)
[Next page](https://discuss.criticalfallibilism.com/t/chatgpt-and-current-ais-are-dumb/1753.md?page=4)