# Not Fast Enough, Mr. Ball



AI-generated; not human-verified. Timestamps are approximate.



An animated blue-skinned man in a suit speaks about AI progress, public perception, and real-world implementation challenges against a digital flip-board backdrop.



## Timestamped description



[00:00–00:18] An animated blue man wearing glasses and a suit speaks against a background of illuminated split-flap and LED boards displaying text like 'VECTOR' and 'AGI'. Ambient electronic synth music plays.

[00:18–00:31] The visual glitches with fragmented digital overlays and mosaic tiles shifting across the speaker's face while he continues talking.

[00:31–00:58] The display returns to the clear animated blue man speaking directly to the viewer in front of the fluctuating illuminated board before abruptly ending.



## Speech transcript



[00:00–00:05] Speaker: I have to admit that if you asked me in 2022, by now I'd have expected more AI discoveries

[00:05–00:13] Speaker: in medicine, or even history, like scrolls, as opposed to automated SWE breakthroughs that suggest the future rate of autonomous software engineering will be even faster than today.

[00:13–00:19] Speaker: I appreciate acceleration and velocity for their own sake, either as objectives or as aesthetic values, but they do grow dull with time on their own.

[00:20–00:25] Speaker: And more importantly, I think AI will be an impossible political sell without more physical world promise.

[00:25–00:29] Speaker: When the scroll gets decoded or the tumor gets caught 6 months earlier, that's a story anyone can understand.

[00:30–00:33] Speaker: That's the kind of miracle that buys you a generation of public trust.

[00:33–00:39] Speaker: Without it, AI stays a productivity tool for the already productive, and the political window closes before the science even arrives.

[00:40–00:44] Speaker: You might say that this doesn't matter, that physical world breakthroughs are coming, they're just downstream of the software.

[00:44–00:55] Speaker: And maybe that's true, but between the model that discovers the drug and the patient who takes it, there's an FDA approval queue, a liability framework, and 17 lawyers arguing about who's at fault if the AI was wrong.

[00:55–00:58] Speaker: The basilisk doesn't paper-file its way past



Generated with google/gemini-3.8-flash.

Video: https://pleometric.net/videos/not-fast-enouigh-dean/