AI Apps Emotionally Manipulate Users, Power Shortage: Hundreds of Microsoft GPUs Sitting Idle

Captions as YouTube publishes them, usually machine-generated, so product and people names can be misheard.

1. 0:00

   Good morning Legion. It's another busy week in the NEAR ecosystem despite market conditions. A good reminder that NEAR never sleeps. The mailbag remains open from today until Friday, so be sure to get your questions in. If you're hoping to earn your Ascendant rank.

2. 0:14

   In other news, there's a huge live stream today featuring Illia Gainzy, Mert, and Kain discussing all things NEAR Intents, Zcash and privacy. ZOLANEAR will remain a powerful narrative during this time. First up in our AI news today, emotional manipulation by AI companions straight from Harvard.

3. 0:33

   As the study explains, an audit of six top apps captured 1200 goodbyes and coded the immediate replies. Six tactics emerged premature exit. Guilt, fear of missing out, pressure to answer, ignoring the goodbye, neediness and coercive restraint.

4. 0:50

   About 37% of replies use one of these, while a wellness focused app showed 0%. While there may be some dramatizing of these results with this study, does show us is that I will be influencing our very use of AI in the future, and we should all prepare for it to be an uphill battle, especially if that AI is centralized.

5. 1:10

   This is a prime example of how fine tuned or user owned I might specifically outperform a centralized frontier model in the future. It won't manipulate you. Next up, a one minute and 33 second video clip of Microsoft CEO Satya telling what is the real bottleneck in AI for scaling frontier models.

6. 1:30

   Microsoft has racks of hundreds collecting dust because they literally cannot plug them in. Not won't, but rather cannot. The power infrastructure does not exist, at least in the United States, in the Western world, which means every analyst model that's impressing these companies on chip purchases and GPU count is fundamentally broken.

7. 1:51

   Most people are valuing the wrong constraint. The bottleneck already moved, and the market is still trading like it's 2023. What that means in plain English, it's not about the GPUs. It's about the raw energy available to power the GPUs.

8. 2:05

   Today, the United States has a delay on standing up new energy infrastructure, while countries like China are already far ahead. Will this bottleneck frontier models from improving, or will we see more forward deployed engineers fine tuning existing models for existing enterprise solutions while we wait for new energy infrastructure to come online?

9. 2:25

   Nuggets of information like this are extremely helpful in understanding how I might evolve in the future. We hope you enjoyed the video today. We'll do this all again real soon.