Post B1wVGYCzSPVAEdTsQK by Phosphenes@mastodon.social
(DIR) More posts by Phosphenes@mastodon.social
(DIR) Post #B1vJKJQ6zMfGv6gZKi by emilymbender@dair-community.social
2026-01-03T20:35:45Z
0 likes, 0 repeats
Not sure why this dross, dated Dec 1, seems to be circulating now (and why it didn't cross my feed a month ago), but wow what a terrible essay.https://bigthink.com/the-present/the-rise-of-ai-denialism/A few comments, in a short 🧵>>
(DIR) Post #B1vJKKSH8nzQ86hoLQ by emilymbender@dair-community.social
2026-01-03T20:35:51Z
0 likes, 0 repeats
First, it displays a certain kind of intellectual laziness of a type I've seen before: Arguing against "pundits", "influencers" and "voices", without naming a single person whose specific arguments the author is arguing against. >>
(DIR) Post #B1vJKLAEVN9SKRGurQ by futurebird@sauropods.win
2026-01-03T23:40:07Z
0 likes, 0 repeats
@emilymbender "By any objective measure, AI continues to improve at a stunning pace."No one ever said "The iphone is improving all the time any moment now everyone will want one"I mean... WHEN will we get there? This "it's getting better" line is a huge tell for me. I do think there is a danger of just hoping AI goes away, it won't. We will need regulations and to think about its place. When you bring that up this kind of guys says "nooo no... it will get better. Please no!"
(DIR) Post #B1vJKSIHzoksRS5wtE by emilymbender@dair-community.social
2026-01-03T20:36:02Z
0 likes, 0 repeats
This allows Rosenberg to caricature the position he is refuting while making it hard for the reader to dispel the caricature, since the original is not pointed to.>>
(DIR) Post #B1vJKabN6klyCx5IkC by emilymbender@dair-community.social
2026-01-03T20:36:10Z
0 likes, 0 repeats
Second, he cites Ayn fucking Rand with (apparently) a straight face. >>
(DIR) Post #B1vJKijSsY03QMvsTA by emilymbender@dair-community.social
2026-01-03T20:36:42Z
0 likes, 0 repeats
He also conflates the form of language (and art) with the actual thing: "produce content" as if output for which no one has accountability (and which represents no one's communicate intent) has any value in the world.>>
(DIR) Post #B1vJKqyeDytkzm67FI by emilymbender@dair-community.social
2026-01-03T20:36:51Z
0 likes, 0 repeats
Experience is central -- no art has any "qualitative value" without experience. Now, people can attribute meaning to synthetic images, but that is also an experience. But as UW's Gabriel Solis once put it so well: writing, art, performance -- these are ways of being human *together*.>>
(DIR) Post #B1vJKytGnrLlXPdvyS by emilymbender@dair-community.social
2026-01-03T20:36:59Z
0 likes, 0 repeats
The words, the pixels, the sound waves aren't the art. The art is in the experience of the artist and the audience, together.>>
(DIR) Post #B1vJL6hrk8ytmMMwJE by emilymbender@dair-community.social
2026-01-03T20:37:31Z
0 likes, 0 repeats
Meanwhile, Rosenberg also consistently displaces accountability: "AI" hasn't done anything. Companies have amassed large collections of stolen art and used them to produce systems with which they and others can create synthetic images.>>
(DIR) Post #B1vJLFJjjE1mU89IRs by emilymbender@dair-community.social
2026-01-03T20:37:41Z
0 likes, 0 repeats
Other companies, who would have previously hired artists, are taking advantage of cheaply produced synthetic images which are only cheap because they are based on stolen art + heavily subsidized by venture capitalists sustaining this bubble.>>
(DIR) Post #B1vJLNUfKTWVqLK8au by emilymbender@dair-community.social
2026-01-03T20:37:48Z
0 likes, 0 repeats
And also unwittingly subsidized by ordinary people paying higher electricity prices when data centers move in:https://www.cnbc.com/2025/11/26/ai-data-center-frenzy-is-pushing-up-your-electric-bill-heres-why.html>>
(DIR) Post #B1vJLVLO8qr8Bt2qFE by emilymbender@dair-community.social
2026-01-03T20:38:14Z
0 likes, 0 repeats
Meanwhile, note the logical leaps here: People creating art based on their experience of others' art is not equivalent to what happens with the large image models munging pixels together. And what is the evidence that "AI" systems will have "sparks of inspiration"?>>
(DIR) Post #B1vJLdUBs0eNRVDz4i by emilymbender@dair-community.social
2026-01-03T20:39:01Z
0 likes, 0 repeats
And finally, this argument about "denial" is really an argument for defeatism (in everyone else, Rosenberg seems to be welcoming the future he paints).>>
(DIR) Post #B1vJLlUq5TvGHpacCG by emilymbender@dair-community.social
2026-01-03T20:39:20Z
0 likes, 0 repeats
If Rosenberg had actually cited the people he's arguing against, and made it possible to click through and see their words, I'm pretty sure what you'd find is not denialism, but resistance.It is important to remember that the future is not yet written. The arts, journalism, science, education, medicine, and our chances to be human together are all worth fighting for --- and not surrendering to the maw of Big Tech./fin
(DIR) Post #B1vJU8wmmXerKnelJQ by swart@cosocial.ca
2026-01-03T23:41:55Z
0 likes, 0 repeats
@futurebird @emilymbender I remember my brother saying to me in the late 80s that "computers still aren't very good" and one day they would be. But they still aren't very good, and he's built his career around them.
(DIR) Post #B1vJxb9L0MfxCVZZOC by ApostateEnglishman@mastodon.world
2026-01-03T23:47:12Z
0 likes, 0 repeats
@futurebird @emilymbender I can't get better though because it's poisoning its own dataset (the internet), with slop. "AI" content has already surpassed human-created content, and that's being fed straight back into the models. So model collapse is inevitable.(I realise you know this already, just underlining your point here!)
(DIR) Post #B1vK36sUT2PT3IJwbw by evan@cosocial.ca
2026-01-03T23:48:15Z
0 likes, 0 repeats
@futurebird @emilymbender I don't understand what you're talking about.ChatGPT has 800M weekly active users. It is one of the fastest growing products in history.https://openai.com/index/1-million-businesses-putting-ai-to-work/The company has revenue around $10B/year.https://www.cnbc.com/2025/06/09/openai-hits-10-billion-in-annualized-revenue-fueled-by-chatgpt-growth.htmlThe product works for people. They use it and benefit from it. It's not "any day now". It's now.
(DIR) Post #B1vKKJgg5xxBhe04ps by futurebird@sauropods.win
2026-01-03T23:51:23Z
0 likes, 1 repeats
@emilymbender Where is AI experiencing natural BIG success?* To generate spam (fake books, videos, "content")* As marketing for stocks (eg our company will make money we use or do AI)* In some narrow science applications* A small segment of the public seems to become addicted to it, in some cases in maladaptive ways. have I missed anything? I know there are many other niche examples, but who will be the main customers?
(DIR) Post #B1vKmMgoWiLV7eREyu by futurebird@sauropods.win
2026-01-03T23:56:26Z
0 likes, 0 repeats
@evan @emilymbender I didn't say "By any objective measure, AI continues to improve at a stunning pace." that was Louis Rosenberg
(DIR) Post #B1vKuatBcJH9Ti9K4W by evan@cosocial.ca
2026-01-03T23:57:52Z
0 likes, 0 repeats
@futurebird @emilymbender right. I think you're trying to make the point that AI is not useful for people? Or am I reading you wrong? It doesn't seem to be borne out by the usage numbers. Maybe you're making some different point.
(DIR) Post #B1vLAk0ZToxYOUIr3Y by Heidentweet@todon.eu
2026-01-04T00:00:48Z
0 likes, 0 repeats
@futurebird The military, like the Israeli Lavender program?@emilymbender
(DIR) Post #B1vLBNQouA84fBsrsu by futurebird@sauropods.win
2026-01-04T00:00:54Z
0 likes, 0 repeats
@evan @emilymbender When I point out the shortfalls someone always says "but it's improving" -- these improvements do not address the shortfalls ... the gap between what is promised and what it really can do.
(DIR) Post #B1vLlmw3YlBYBerBQG by evan@cosocial.ca
2026-01-04T00:07:31Z
0 likes, 0 repeats
@futurebird @emilymbender oh, ok. Welp, hundreds of millions of people are using ChatGPT on a weekly basis right now, despite whatever shortfalls you're pointing out. And that's just one product. So what it really can do, right now, is scratching a lot of people's itch.
(DIR) Post #B1vM2D0APhLpfCbqAC by futurebird@sauropods.win
2026-01-04T00:10:31Z
0 likes, 0 repeats
@evan @emilymbender A lot of people use blockchain too every day. It has applications.
(DIR) Post #B1vMa6uVZDnQcM80Aq by evan@cosocial.ca
2026-01-04T00:16:33Z
0 likes, 0 repeats
@futurebird @emilymbender it does!
(DIR) Post #B1vNu4DfNjA819AXxI by DrHyde@fosstodon.org
2026-01-04T00:31:26Z
0 likes, 0 repeats
@futurebird @emilymbender analysis of Go and Chess games!
(DIR) Post #B1vWz7E2X2mINMpGca by graydon@canada.masto.host
2026-01-04T02:13:10Z
0 likes, 0 repeats
@futurebird @emilymbender The vaguely-sane financial press puts the current direct capitalization for AI at somewhere around one point five trillion USD.Assuming the folks using this cash flow to skim feed are managing to retain about ten percent, that's a hundred and fifty billion dollars concentrated in grifter pockets.It also sets up major economic failure, thus further cementing incumbency, weakening the civil power, and expanding the scope of rent.All are major grifter successes.
(DIR) Post #B1wPduQOQONE4ZVMGG by apophis@brain.worm.pink
2026-01-04T02:36:39.465566Z
1 likes, 0 repeats
@futurebird @emilymbender mass surveillance?
(DIR) Post #B1wPdveFsL3XrqzwVU by futurebird@sauropods.win
2026-01-04T12:25:39Z
1 likes, 1 repeats
@apophis @emilymbender I can't tell if it's really useful for this, or if that's just marketing. All of the worst people have been told that these tools will let them predict what the public will do next. It will let them find the key radicals and dismantle their networks. Perfectly inject propaganda and cash to take down their enemies. And the chat bot will *say* I can help you make a plan cheerfully enough. But ... is it going to be a good plan? These people expect magic.
(DIR) Post #B1wQPPaqMHbILo2mv2 by futurebird@sauropods.win
2026-01-04T12:34:15Z
1 likes, 0 repeats
@apophis @emilymbender There are many gullible rich people who think they can use AI to control public opinion, to gird the loins of their already massive power. I tend to think this will fail spectacularly and hilariously. But... IDK talk me out of that if you can.
(DIR) Post #B1wVGYCzSPVAEdTsQK by Phosphenes@mastodon.social
2026-01-04T13:28:36Z
0 likes, 0 repeats
@futurebird @apophis@brain.worm.pink @emilymbender AI is good at surveillance and deception. Not the high level stuff, but if you want to turn 1 million phone calls into searchable text, you've got it. If you want to categorize millions of pages of searchable text into "dissent?: yes/no", it won't be as reliable as humans, but no human could process such volume.As for deception, maybe it won't fool an idiot but it might confuse one. That compromises maybe 33% of USA, enough to win any election.