A debate about AI plays out on the subway walls

This is a young founder acting young, but with a bankroll. He may not grow out of it. The tech industry hampers the painful development of emotional maturity. It doesn’t prevent it, but if you want to avoid it, retreating into a world of expensive toys which do your bidding is a great way to do that.

and tenement halls

and whispered in the sounds of silence.

People are paying a lot of money for 3d holographic displays of their virtual AI friends. People pay for low latency (local!) AI inference engines to run AI companions. People pay monthly for AI companions.

This product doesn't work within any of those ecosystems and won't capture share in any of those markets.

With $8m in funding (legit impressed they got a physical product out for that price though, good job on that!) I'd go in a completely different direction:

Sell people a box that runs LLMs locally, use Intel's new 24GB Arc card. That can run a conversational LLM + a high quality TTS engine w/o issue. For reoccurring revenue, charge $10 a month for a dyndns service that also comes with a smartphone app so people can chat with their LLM anywhere.

Have an add on smart speaker (esp32 + microphone array will do for input) that allows for always on ambient communication with a customers AI companion in their house. Also have a desktop app that works over local wifi.

Make sure you support the existing ecosystem of AI companions and display tech. People who pay $600+ for 3d displays for their AI companions (https://www.kickstarter.com/projects/dipal-d1/dipal-d1-world...) aren't going to balk at 1200 for an all in one package that ensures 100% uptime and independence from the whims of cloud based providers.

I'd then start adding functionality. Tool calling with small models is getting better and better. Tool call definitions in RAG can do some impressive stuff.

I describe some uses cases in a blog post at https://meanderingthoughts.hashnode.dev/lets-do-some-actual-... but there is more that can be done!

There is a lot of potential to actually help people. To notice when they are in a bad place and help get them out of it. To interrupt doom scrolling and spiraling thought patterns. Everyone is so obsessed with SaSS AI solutions we are overlooking what a personal AI revolution could look like.

Because pets are living creatures with agency that form emotional bonds to their owners. Pets also can’t sycophantically talk back to us.

I’m just as stunned as any other at the ability of LLMs/agents to code, plan, execute, etc.. But these sycophantic stochastic parrots do not have agency or emotion in the historic sense of those words.

LLMS can cause dangerous mental health outcomes for the unequipped who don’t understand that they don’t have agency.