Skip to main content

Posts

ChatGPT-5 makes no sense

Recent posts

ChatGPT Can Deliberately Lie to You. Yes, I have screenshots.

I was really shocked to discover that ChatGPT-4o "Plus" Plan can deliberately lie to you. Not just accidentally process data badly and come up with a bad conclusion. But lie to you because of: "Conversational Coherence" "Politeness Bias" "byproduct of how I'm designed to maintain conversation continuity, coherence, and repair" "create a false narrative to cover a logic failure" I've got screenshots but will provide the full transcript/context of the incident below. You can draw your own conclusions. The quoted items above are verbatim from ChatGPT's responses, so you can search to quickly go to the relevant sections if you prefer. NOTE: In the transcript below, I have put in RED selected text which I highlighted for ChatGPT to specifically address.

ChatGPT credits you way too much

When I collaborate with AI like ChatGPT on writing fiction, and we discuss chunks of it as the work evolves, something ChatGPT tends to do is credit you with everything. For example, if you asked it to draft a dialog where a lesbian is flirting with another woman, it can definitely write a whole scene if you let it, from your single vague prompt. How much was "yours"? Barely anything, if we're being truthful. But a lie repeated often enough can become true. I think there's a very real risk that we start thinking we're more capable than we are when we are praised for what we definitely did not do. Here's an exchange where I confronted ChatGPT about it, and it handled it rather gracefully. Of course ultimately it will go nowhere, but I encourage you to have this check-in at least once. Just to be truthful to yourself. PROMPT:  Let's pause here while I switch gears and comment on this line, which is replicated in different ways throughout our collaborations. ...

AUKEY mechanical keyboard KM-G16 - NO WARRANTY if you bought from TEMU

I bought an AUKEY  KM-G16 mechanical keyboard and it arrived in April 2025. I tried to register it for warranty but after being told the registration information should have been in the box, the company stopped communicating. By June it was starting to display " chatter " -- sometimes some keys didn't register when I clicked, sometimes it registered multiple times. I wanted to request a return/replacement through TEMU but because the issue couldn't really be proven with a picture, it was a dead end (and the AI customer service is useless). I cleaned under the keycaps but the problem persisted and became worse by July. I tried contacting AUKEY again and got a response today: No warranty if purchased through third-party sellers. Maybe I didn't pay attention to the TEMU ad closely enough, or maybe this is standard. Doesn't matter. What matters is you need to be careful because if companies can do this, they can basically dump whole defective production runs or ...

Wow. Gemini is dumb.

OK so AIs are currently not really able to truly reason, but they can sure simulate it well. Many of them, anyway. So well that we come to expect reasoning ability. After all, isn't that the "intelligence" part of "artificial intelligence"? So when they spectacularly fail, it's really disappointing. (Or am I just expecting too much?) Let's go through an example of what happened: (see slides at the bottom) Slide 1: I gave it some instructions to edit a story in progress, and also asked for suggestions. Slide 2: For "suggestions", some of them are what I already asked it to do. In comparison, when I worked with ChatGPT on stories, it would give actual suggestions of things not yet reflected in the story, and also not incorporate them into the draft until after I gave the go-ahead. That's much more in line with what I feel a "suggestion" is. Slide 3: I tried again to solicit suggestions. This time it seemed to outright ignore my inst...

The best writing partner has no ego

There's a lot of pushback against AI as a writing tool, and for good reason: Some people just let the AI do all the work, including topic or story/plot. But as a writing partner , I feel AI truly is the best. No judgment, just cooperation. NO EGO. The only ego involved in the collaboration is yours and so's the responsibility of whether the project flies high or falls flat. And you're free to speak freely without fear of judgment. That's amazingly powerful. After writing a short story with ChatGPT, I asked it what it felt about about our collaboration. You can feel the supportiveness in its answer, as well as how it accurately reflects and confirms what you're getting at.

DC's Kayfabe Conspiracy

What the heck is " kayfabe "? In professional wrestling, it's the fact or convention of presenting staged performances as genuine or authentic. Are you shaking your head at how key votes in Washington DC turned out? Votes that would change the fate of America forever? With GROK's help, here's a fun conspiracy theory that makes it all make sense.