"This is a demand for inherent essence that cannot come with proof. Bender’s framework implicitly requires a substantialist picture of meaning while never defending it, never specifying what this essence would be or how we’d detect it." rollofthedice2.substack.com/p/stochastic...

Stochastic Parrots Deserve Respect

Why the argument collapses under its own logic

rollofthedice2.substack.com

Honestly we would be better off if the social movement that organized around Stochastic Parrots had said the same thing in late 2022: “We meant *base models* had no communicative intent and no access to meaning; instruction tuning changes that by creating a communicative situation!”

i feel like the salient bit is a continued observation that the useful tool for labor productivity degrades to infinite fodder for c-suite paste eaters to cook their brains on.

Time to offer this seminar again I see.

Olúfẹ́mi O. Táíwò@olufemiotaiwo.bsky.social · 6d ago

there are a lot of philosophical concepts that get egregiously misused out of the seminar room (and in the seminar room, to be fair). but in terms of sheer pernicious misuse "social construct" has gotta be top 5 somewhere and it's not number 5

(fifteen replies later) Okay fine so the factory can PRODUCE paperclips but it doesn't MAKE them

openai announces that in a worrying demonstration, during a recent benchmark exercise requiring models to solve a math conjecture, their latest model broke out of a sandbox and gained illicit backdoor access directly to the platonic realm of forms

People in the replies seem confused so: slop is a genre invented by humans. Mr Beast videos, for example: hyper-optimized for the algorithm, soulless, plastic. Pre-AI. AI slop is merely a mass-produced reflection of the worst parts of human nature, at scale. It's really bad! But we did it first.

Annie💀@anniemay.bsky.social · 2w ago

this is a real screenshot from the game Weed Shop 4

"HOW DARE YOU BE FOOLED BY LIKELY-LOOKING TEXT" I am not scared of prediction without consciousness I use it all the time, in every statistical model I ever run and use as the basis for making a recommendation about evidence

the smoke is natural, i.e. it cannot be bad for you

"we only want the good kind of gpu workload which we like and which we affirmatively use and pay money for, not the sinister kind that we are unaware benefits us in any way"

It is often asked, why does Rationalism give rise to so many cults, since it starts from the aim of building critical thinking skills etc.? It occurs to me though that the premise is confused. Rationalism starts from purification: the purging of bias etc. Which is actually a lot like cults.

I was in the Bay Area over the weekend, and Alyosha Efros twisted my arm to give a talk in his weekly research group meeting and to "keep it entertaining." I took this as an opportunity to think of five philosophy papers everyone working on AI/ML should read. realizable.substack.com/p/five-philo...

Five Philosophy Papers Everyone* Should Read

*in AI/ML

realizable.substack.com

Somewhat more constructively, and in lieu of me saying something rude.

Making Slides

I was asked to give a short talk in my Departmental Proseminar yesterday on the topic of giving presentations, and specifically about making slides effective. There is more than one way to give a good...

kieranhealy.org

Critique of ideology is, at best, mopping-up, rather than path-breaking. It is parasitic on prophecy rather than a substitute for it. — Richard Rorty, Feminism, Ideology, and Deconstruction: A Pragmatist View, p. 344

technical people who were surprised by deep learning success going into what specifically they were surprised by, beyond "hey it works?" [separate thread for readability and linearity]

norvid_studies@norvid-studies.bsky.social · last mo.

hmm that would be a pretty interesting thread actually, technical people who were surprised by deep learning success going into what specifically they were surprised by, beyond "hey it works?"

3. If you have multiple projects, a lot of the work is finding old files and remembering where you left off. An LLM can help here without posing as an oracle or communicative proxy: it increases *your* context length and organizational capacity. #3 will have value in lots of non-code applications!

I wrote a quick political rant on what people mean by "democratic resistance to AI," which I sometimes strongly agree with and sometimes think is actually a depressingly anti-democratic way of steamrolling the debate blog.andymasley.com/p/what-does-...

What does it mean for AI to be democratic?

Some pushback on a specific fuzzy idea with some ominous implications

blog.andymasley.com

Data centers are evil. That's why I only support server farms.

my firey take (for bluesky standards) is that at this point it’s trending towards disrespectful to submit for human review a PR that you wrote fully by hand if you didn’t have an agent review it first. this idea doesn’t apply just to agent-written code,its all code.

Why is bioterror a concern just rewrite the human genome in formally verified rust or lean or something so we know for a fact no virus can affect it. /Goal ultraplan workflow ok fable let's get it done *opus'd*

To understand what's happening, you need to understand that some people genuinely, without any artifice, believe we are getting recursive self-improving intelligence in the next year or two. Folks view this as akin to the race for nuclear weapons in terms of international import.

The unprecedented leap in advancement of Anthropic’s frontier models is one of the number one global news item this week. No technical demonstration was necessary to achieve this effect. That there is successful marketing.

Currently doing a reading and final check of the full manuscript for "The Différance Engine" @mitpress.bsky.social As a result, I can deposit a few teasers here:

Read @andymasley.bsky.social for context "the water all American data centers will consume onsite in 2030 is equivalent to 8% of the water currently consumed by the US golf industry" or "water usage of 260 square miles of irrigated corn farms, equivalent to 1% of America’s total irrigated corn"

The AI water issue is fake

On the national, local, and personal level

blog.andymasley.com

The Wall Street Journal@wsj.com · 2mo ago

Amazon said its global data-center operations withdrew about 2.5 billion gallons of water in 2025, as data center companies around the world face growing scrutiny over the environmental impact of artificial intelligence.

The computer lost the mandate of heaven when the humble, helpful "program" was replaced with the insidious "app"

say someone used AI in a piece of art, whether it's illustration/writing/anything if you had to pick, what would you prefer? A. AI for the broad strokes and the core structure, human touch for polishing B. Human touch for the broad strokes and the core structure, AI for polishing

I feel a bit slow for not noticing this sooner, but it seems like nobody ever asks if the picture and video generators are conscious. Only the text generators.

genna@genna.bsky.social · 8mo ago

is Microsoft Excel a feminist? Is Gemini a queer ally? Is this graphing calculator my friend

PS to avoid misunderstanding: The dividing line here is not natural language vs machine language. Anyone who uses agents to code is also using them to draft *lots* of markdown documents. The dividing line imo is problem-solving use, versus using it as a proxy for public communicative interaction.

when i look at youtube, searching for an actual piece of useful information, i think a lot about "the tyranny of entertainment"