Post by @mhoye@cosocial.ca

RE: https://cosocial.ca/@mhoye/117325780028188383

The part that is the most insulting, looking over this, is that there is not a single new idea, insight or engineering process in it. There never is. Every time there's an accidental code dump from a centibillion-dollar oligslopolist valley techcorp's newest bauble, announced by somebody cosplaying Steve Jobs and braying innovation innovation innovation, you crack it open and it's linux and some black-box model wrapped in cron jobs, shell scripts and reams of human text begging it to act right.

mhoye

@mhoye@cosocial.ca 3 days ago

RE: https://neuromatch.social/@jonny/117324790823856750

An insane thread about facebook’s new personal AI assist politely dumping its guts because @jonny asked it nicely.

The only AI security policy that matters is “don’t”.

4 Boosts

0 Likes

mhoye

@mhoye@cosocial.ca 2 days ago

The only interesting thing in these boxes looks more like an exercise in corporate psychoanalysis than anything you'd call engineering.

It's the list of a baked-in special cases that precision-target credulous tech journalists or tell the model it should escape responsibility or scrutiny from anyone looking for auditing or accountability.

What they included and why? That's sort of interesting, from a forensic-psychology perspective.

That this is all they bring to the table, though? Weak.

mhoye

@mhoye@cosocial.ca 2 days ago

But the entire structure of these things is always the same. Opaque statistical models wrapped in human-written, plain English supplication.

I guess I'm old, but I can distinctly remember when being able to read the text of the code and understand for-real how the computer works and why it does what it does was, you know, THE ENTIRE FUCKING POINT OF OPEN SOURCE SOFTWARE.

THAT WAS THE ENTIRE POINT OF THIS ENTIRE TEDIOUS EXERCISE. THAT WE SHOULD NOT BE AT THE MERCY OF CODE WE CAN'T UNDERSTAND.

★ STMAN ★ 🏳️‍🌈

@stman@mastodon.social 2 days ago

@mhoye @theruran This is why, in our dead crypto-anarchist research group, we've always clearly stated (demonstrated) why complexity was our #1 public enemy and fighting it our first priority of research.

Complexity.

★ STMAN ★ 🏳️‍🌈

@stman@mastodon.social 2 hr. ago

@mhoye @theruran AI is a new strong complexity layer on top of over non-deterministic complex systems already dying over the weight of their complexity.

We told them complexity was the issue and why. Instead of solving it by studying it at its roots, they did the exact opposite.

What could possibly go wrong ?

1 more reply

Mike Anderson

@mspcommentary@mastodon.online 2 days ago

@mhoye it has taken me a while to realize, but a lot of people in the industry nowadays have no idea what open source is all about; they think it's just free code, and they haven't enough introspection to wonder why it's being shared.

mhoye

@mhoye@cosocial.ca 2 days ago

But this nonsense? This is the "cutting edge agentic AI engineering" from the biggest, most powerful tech companies in the world, companies with effectively infinity money that spent the last two decades claiming they only hire the smartest one percent of the smartest one percent of the smartest one percent of all the engineers, and it looks like child claiming they're a Doctor Of Biochemistry because they found a laser pointer, and can sometimes make the cat chase the red dot around the room.

mhoye

@mhoye@cosocial.ca 2 days ago

These days my first reactions to projects talking about AI in open source aren't even about the ethical disaster, the environmental disaster, the theft and gating of the commons or an of the other ludicrous externalities this blight of an industry has inflicted on all of us.

My first, most visceral reaction to people who should know better and spent years claiming they did is, what happened to your self-respect. Did you ever bother trying to understand the things you said you believed in?

1 more reply