Reckons

Reckons

119 bookmarks
Newest
You're Supposed To Be Glad Your Tesla Is A Brittle Heap Of Junk | Defector
You're Supposed To Be Glad Your Tesla Is A Brittle Heap Of Junk | Defector
Tesla cars are shoddily built pieces of shit liable to fall apart and malfunction in dangerous ways at inopportune moments. No, this is not a blog from 2012! It is also not a blog from 2015 or 2018 or 2022. It is not even a blog from two weeks ago about Tesla’s self-driving systems killing […]
The reason Tesla hasn't "worked all the bugs out yet" is that the company is run by people who hold established best practice in ideological contempt
In the absence of any real opportunity to envision a brighter future, you sign up to support some inheritance goober's personal fantasy camp by dumping money into his company or buying his stupid-looking car. In this way, you are meant to understand, you have participated in the great grand adventure of discovering tomorrow.
You're Supposed To Be Glad Your Tesla Is A Brittle Heap Of Junk | Defector
A Camera, Not an Engine
A Camera, Not an Engine
Modern AI puts us firmly into an age of exploration of computational reality
But why stop with datasets that induce languages with “grammars” that can be rendered legible to us? Could you make a “Large Solar Flares and Sunspots Model” (LSFASM) and learn to talk to the Sun and ask it where it might flare up next? How about a Large Oceanic Model that allows ships to talk to ocean currents? Or a Large History Model that works as a Prime Radiant for Asimovian psychohistory? Maybe a Large Climate Model constructed out of weather data can talk to us and supply strategies for climate change?
One reason it is hard is, once again, our tendency to mistake discoveries for inventions, or equivalently, cameras for engines. Instruments of discovery measure more than they are measured. Yes, there are a number of ways you can measure a telescope (mirror diameter or focal length for example), but the interesting measuring going on is what the telescope is doing to what it’s turned towards (the analogy to AI here is perhaps to things like floating-point precision — that’s closer to mirror diameter).
A Camera, Not an Engine
Fire and Motion
Fire and Motion
Programming Note: Summer Lightning will be on a break till the new year. Thanks to everyone who has supported my writing in the last year. This is the 57th and last edition of the newsletter for 2023. I will pause paid subscriptions, which will resume in January 2024. You can find the archive of posts
Fire and Motion
Unbundling AI — Benedict Evans
Unbundling AI — Benedict Evans
ChatGPT and LLMs can do anything (or look like they can), so what can you do with them? How do you know? Do we move to chat bots as a magical general-purpose interface, or do we unbundle them back into single-purpose software? What are the products?
Unbundling AI — Benedict Evans
Introducing public design
Introducing public design
Andrew Knight, Head of UK Policy Design Community, invites designers to participate in the Public Design Review.
The absence of a common proposition divides rather than unites the design family, and it is the number one barrier to bootstrapping design in governments. We need a universal definition and proposition for design in the language of governments and ministries of finance.
Introducing public design
Elegant and powerful new result that seriously undermines large language models
Elegant and powerful new result that seriously undermines large language models
Wowed by a new paper I just read and wish I had thought to write myself. Lukas Berglund and others, led by Owain Evans, asked a simple, powerful, elegant question: can LLMs trained on A is B infer automatically that B is A? The shocking (yet, in historical context, see below, unsurprising) answer is no:
Lukas Berglund and others, led by Owain Evans, asked a simple, powerful, elegant question: can LLMs trained on A is B infer automatically that B is A? The shocking (yet, in historical context, see below, unsurprising) answer is no
Elegant and powerful new result that seriously undermines large language models