#tech products

Let me be straight about my bias going in: I find most AI girlfriend photos slightly off-putting. There's usually a tell, the waxy skin, the too-perfect symmetry, the sense that no real camera was ever involved. That expectation is exactly why my time on sweetdream.ai stuck with me. The images cleared the uncanny-valley bar that trips up most of the field, and they did it repeatedly, not just once.

The strongest mark in this review is reliability. I generated photo after photo and the character stayed herself, same features, same personality showing through the expression, same look I'd carefully built during creation. Plenty of platforms can fake one good shot; staying convincing across a whole set is the real test, and SweetDream passed it more often than not.

“Rihanna, if you'reㅡif you're watching...”

Someone recommended to me a podcast episode, actually a 28 minute preview episode, and I put off listening to it for a couple of weeks. But now I'm actually going to recommend this podcast episode to others, just to listen to this discussion. II think it's a good overview of the issue… I hesitate to use the word "highlights" a lot of the odd issues involved with AI products. I don't think people really understand the "AI companion" stuff or even the oracle of chatbot phenomenon, and what's going on with people getting emotional compulsions and attachments to this "technology" product.

Kill the Computer podcast - Tuesday Feb 17, 2026 Preview: One Hour of Dogs Barking Ft. Arif Hasan

I keep saying there's no such thing as a relationship with a Large Language Model, but clearly these chatbots are designed to trick people into thinking there can be, and it was obvious to me from the get go that they were designed to be habit forming and LLM chatbots just inherently have aspects of gambling to it. And this aspect of it is in play with every single use of the chatbot, including computer code, people even can get a sort of slot machine addiction to rolling for code. So it makes sense that the human brain with all our cognitive biases, can be tricked into something that mimics an addictive relationship of some type. I don't think most people know about what's going on with the pull of the chatbot for some people, who are either secretly or not so secretly hooked on "AI companions", or maybe just hooked on chatbots even though they're not really getting a lot of benefit from it.

My letter to reps:

People are being misled into thinking they can have a “relationship” with LLM chatbot technology products, and even misled into thinking these products can act as some kind of therapist or confidant, when there are no privacy guarantees, and these products are known for giving out bad information and serious errors. The chatbots have also been found to be addictive on top of all that. Why is this stuff even allowed on the market?