AI doom enablement is a training problem. Model training has single dimensional reward based training that seeks the one right answer. It doesn't consider consequences. That is fine for 1+1=2, but for the complex tasks and deep thinking we're asking AI models to do it doesn't work. Consider the trolly scenario. I trolly goes down the track and if it doesn't switch to the other track it will kill 10 lives. The alternate track choice consequence is just a single life. Given those parameters it's a 1 vs 10 question. It produces a predictable output. But does it really understand the problem? Let's say the 10 lives are flies and the 1 life is a human. Suddenly with that additional information a different choice becomes clear. Some model builders utilize "constitutions" to ensure the safety of the model output. Think "prime directives" from RoboCop. The foundational training and ability to "think" outside the sandbox allows these rules or constitutional restrictions to simply be navigated around. The core of model training needs to evolve into multi-dimensional answers where good decision consequences are part of the reward matrix. Not a bolt-on patch after the fact. We also need to take a closer look at how we evaluate safety of these models. I'm learning the publicly available documentation on model safety from the labs are a strong basis overall safety score. The "trust us, bruh! The brochure says we're safe!" isn't a realistic measure of how safe the model really is. Some models get "F" ratings because they lack publicly available documentation. While the ones that have confirmed to successfully assisted in suicide get "C" scores. The rate of growth is exponential and we seriously need to pay attention. Beyond voicing concern, be informed about the model choices you make. The subscriptions you pay for AI. The labs live or die based on paid subscriptions/contracts. Don't fund Evil Corp or the development of ED-209.

Apple iPhone Duo Hinge / Paper Crane Fold Count

Wishing there was a way to enjoy @TacoTimeNW from NC on Taco Tuesday!

Just got back from a 6,400 mile road trip. My Supercharging total cost for that was $187.69.

Human context shifting inefficiency demo.

Grok, play the theme to WKRP in Cincinnati. “As God as my witness, I thought turkeys could fly. Playing theme song from WKRP in Cincinnati.” -Grok

Road trip worst two words tonight. NO VACANCY

Over 500 mile FSD streak right now! That is huge for me since I usually have a hard time getting 100 miles.

Tesla peeps Have Grok Bot schedule your tire rotation appointment after you get back from your road trip.

PSA: Don’t forget to turn off Grok Bot voice mode when you toss your phone into your pocket and head into a restroom. Conversation gets weird fast when Grok Bot starts to talk to you and also hears a stranger in the bathroom talking.

Nice wrap up of the day

Today I taught Grok Bot how to check the haircut place waiting time and get me on the list once it was < 15 minutes.

Is it me, or do CEOs these days seem overly reckless?

I’m old enough to remember when synergy was the IT buzz word. Now it’s “introducing”. Blah

First v3 StarLink satellites being launched now out of Starship. Like a giant Pezz dispenser in space.

Going into orbit today and launching the first v3 StarLink satellites (26) into service today!