Benjamin Harnett
@benjaminharnett.com
Sometime poet, historian, software engineer, novelist, union man. https://www.benjaminharnett.com https://thehappyvalleynovel.com
Pinned
Benjamin Harnett
@benjaminharnett.com
· Jun 20
I interviewed 300 high achievers about their morning routine, and you will never believe, they all have inherited family wealth.
Reposted by Benjamin Harnett
Again and again, we see higher ed institutions ignoring and rejecting the collective expertise of their faculty and staff 🫠
This decision was made without consulting a single CU Boulder professor with AI expertise. And AFAIK there was one professor *total* involved.
More info, such as it is, here: https://www.cu.edu/gen-ai
Check the "Guiding Principles" section for entertainment.
More info, such as it is, here: https://www.cu.edu/gen-ai
Check the "Guiding Principles" section for entertainment.
February 11, 2026 at 10:51 PM
Again and again, we see higher ed institutions ignoring and rejecting the collective expertise of their faculty and staff 🫠
Reposted by Benjamin Harnett
It’s possible AI can do some things well but it doesn’t matter because in order for those businesses to still exist in two years, it needs to do everything
February 11, 2026 at 8:10 PM
It’s possible AI can do some things well but it doesn’t matter because in order for those businesses to still exist in two years, it needs to do everything
Reposted by Benjamin Harnett
Tonight event is:
February 11, 2026 at 6:26 PM
Tonight event is:
Reposted by Benjamin Harnett
a bunch of people who work for AI companies and largely paid in vesting stock options are overwhelmed by the magnitude and depth of what they've created. everyone else is a curmudgeon or otherwise a confused hater. truly this is an irreconcilable intellectual dilemma.
Experiments conducted with the A.I. system Claude are producing fascinating results—and raising questions about the nature of selfhood. Gideon Lewis-Kraus reports from inside the company that designed it, Anthropic. newyorkermag.visitlink.me/rOfXjg
February 11, 2026 at 6:12 PM
a bunch of people who work for AI companies and largely paid in vesting stock options are overwhelmed by the magnitude and depth of what they've created. everyone else is a curmudgeon or otherwise a confused hater. truly this is an irreconcilable intellectual dilemma.
Delusion as a service
My new favorite LinkedIn AI posts go like this:
1. identify real problem with GenAI that is clearly inherent in the technology and makes it ludicrously unfit for supposed task
2. here's a hack that completely obscures this problem (remember the problem is inherent) so you can use GenAI for the task
1. identify real problem with GenAI that is clearly inherent in the technology and makes it ludicrously unfit for supposed task
2. here's a hack that completely obscures this problem (remember the problem is inherent) so you can use GenAI for the task
February 11, 2026 at 6:09 PM
Delusion as a service
This is mush-brained bullshit that completely obscures what is actually going on with this tech: the hype is not true but the consequences are, and you are directly enabling those consequences by thinking like this
I understand why people are exhausted by AI hype, and why those of us squarely in the corner of "human dignity uber alles" see AI doomerism as self-serving hype, but I *really* think people on the left broadly need to start thinking seriously about the possibiltiy of the hype being...true.
February 11, 2026 at 6:09 PM
This is mush-brained bullshit that completely obscures what is actually going on with this tech: the hype is not true but the consequences are, and you are directly enabling those consequences by thinking like this
My new favorite LinkedIn AI posts go like this:
1. identify real problem with GenAI that is clearly inherent in the technology and makes it ludicrously unfit for supposed task
2. here's a hack that completely obscures this problem (remember the problem is inherent) so you can use GenAI for the task
1. identify real problem with GenAI that is clearly inherent in the technology and makes it ludicrously unfit for supposed task
2. here's a hack that completely obscures this problem (remember the problem is inherent) so you can use GenAI for the task
February 11, 2026 at 6:06 PM
My new favorite LinkedIn AI posts go like this:
1. identify real problem with GenAI that is clearly inherent in the technology and makes it ludicrously unfit for supposed task
2. here's a hack that completely obscures this problem (remember the problem is inherent) so you can use GenAI for the task
1. identify real problem with GenAI that is clearly inherent in the technology and makes it ludicrously unfit for supposed task
2. here's a hack that completely obscures this problem (remember the problem is inherent) so you can use GenAI for the task
Reposted by Benjamin Harnett
2013 was a full 5 years after Epstein was convicted of sex crimes with minors when this email for Hakeem Jeffries was sent.
Gonna need someone to look at the date on this thing because it's uh not great if this is real. It's in the files.
February 11, 2026 at 4:16 PM
2013 was a full 5 years after Epstein was convicted of sex crimes with minors when this email for Hakeem Jeffries was sent.
Reposted by Benjamin Harnett
Also I’m sorry but I don’t think this is Claude developing a personality or ethical values. It is *simulating* those things because that is what humans value and reward. The programmers are falling for their own Clever Hans trick!
February 11, 2026 at 3:47 PM
Also I’m sorry but I don’t think this is Claude developing a personality or ethical values. It is *simulating* those things because that is what humans value and reward. The programmers are falling for their own Clever Hans trick!
Reposted by Benjamin Harnett
Reposted by Benjamin Harnett
The guy has been making promises he can't keep for decades, yet to his followers, and those investing in his company, it just doesn't matter.
He's like the end times guy who, when his latest date for Armaggen passes without incident says, "forgot to carry the three", and gives out a new date.
He's like the end times guy who, when his latest date for Armaggen passes without incident says, "forgot to carry the three", and gives out a new date.
February 10, 2026 at 8:24 AM
The guy has been making promises he can't keep for decades, yet to his followers, and those investing in his company, it just doesn't matter.
He's like the end times guy who, when his latest date for Armaggen passes without incident says, "forgot to carry the three", and gives out a new date.
He's like the end times guy who, when his latest date for Armaggen passes without incident says, "forgot to carry the three", and gives out a new date.
Personally, if I were a prestige magazine like The New Yorker, I wouldn't be doing unpaid PR for an AI company masquerading as informing my readers. But that's just me.
February 11, 2026 at 2:37 PM
Personally, if I were a prestige magazine like The New Yorker, I wouldn't be doing unpaid PR for an AI company masquerading as informing my readers. But that's just me.
Reposted by Benjamin Harnett
I drew a face on a paper plate. Who could even begin to understand how it perceives the world or what it dreams of?
Researchers working with the A.I. system Claude have come to understand that the model’s selfhood, like our own, is a matter of both neurons and narratives. newyorkermag.visitlink.me/kEX1vV
February 11, 2026 at 4:39 AM
I drew a face on a paper plate. Who could even begin to understand how it perceives the world or what it dreams of?
There is one and only one way to change tech companies from the inside, and it starts with unionizing. After that it’s a brutal slog.
“Beiermeister started at OpenAI in mid-2024 as a part of a wave of hires from Meta who viewed themselves as trying to change tech companies from the inside.”
OpenAI fired a VP who opposed their erotica rollout (and had started a company-wide peer mentorship program for women), allegedly telling her it was because she discriminated against a man 🤔 www.wsj.com/tech/ai/open...
February 11, 2026 at 1:20 PM
There is one and only one way to change tech companies from the inside, and it starts with unionizing. After that it’s a brutal slog.
Reposted by Benjamin Harnett
"Only two English words rhyme with culture, and these, as it happens, are sepulture and vulture. We don’t yet call museums or galleries or even universities culture-sepultures, but I hear a lot, lately, about culture-vultures (man must rhyme)."
Raymond Williams
Raymond Williams
February 11, 2026 at 1:16 PM
"Only two English words rhyme with culture, and these, as it happens, are sepulture and vulture. We don’t yet call museums or galleries or even universities culture-sepultures, but I hear a lot, lately, about culture-vultures (man must rhyme)."
Raymond Williams
Raymond Williams
Reposted by Benjamin Harnett
Sometimes I remember Reagan removing Jimmy Carter’s solar panels and I think we’ve been enduring this shit for way too many years.
February 11, 2026 at 6:25 AM
Sometimes I remember Reagan removing Jimmy Carter’s solar panels and I think we’ve been enduring this shit for way too many years.
Reposted by Benjamin Harnett
The TN GOP doesn't want its constituents to know that most of the vouchers they are paying for are going to the affluent who already had their kids in private school and that what they said about what vouchers would do -- give parents "choice" -- was a lie.
Tennessee House panel kills private-school voucher transparency bill • Tennessee Lookout
Tennessee Rep. Jody Barrett is sponsoring a bill to reveal family income of kids receiving state voucher funds to attend private schools.
tennesseelookout.com
February 11, 2026 at 12:20 PM
The TN GOP doesn't want its constituents to know that most of the vouchers they are paying for are going to the affluent who already had their kids in private school and that what they said about what vouchers would do -- give parents "choice" -- was a lie.
Anybody who spends regular time engaged with ai chatbots is in danger of being groomed into a psychotic break and puffed so full of delusions as to be dangerous to themselves and others. I’m looking at you, senior corporate leadership and politicians around the globe!
I will share this with my Bible and Misinformation class. It is an example of why teaching students what actual research involves is more important than ever, and needs to be coupled with explanation of how chatbots reaffirm the bizarre things users insist on.
New evidence shows that 'Jesus was never crucified' – and AI can prove it
Julian Doyle, a veteran filmmaker behind Monty Python's Life of Brian, says he's spent decades researching and claims the man put to death was not Jesus but someone else entirely
www.dailystar.co.uk
February 11, 2026 at 12:22 PM
Anybody who spends regular time engaged with ai chatbots is in danger of being groomed into a psychotic break and puffed so full of delusions as to be dangerous to themselves and others. I’m looking at you, senior corporate leadership and politicians around the globe!
If I’m honest, I’m still chasing this reading experience. Talk about beginning in medias res.
True story: I read the two towers first because fellowship had been checked out of the library when I wanted to read lord of the rings. So you can imagine it was weird for me in general.
February 10, 2026 at 9:23 PM
If I’m honest, I’m still chasing this reading experience. Talk about beginning in medias res.
Reposted by Benjamin Harnett
HOOOOOO boy.
Rep. LaMonica McIver (D-N.J.): "Do you consider yourself a religious man?"
Lyons: "Yes ma'am."
McIver: "How do you think Judgement Day will work for you with so much blood on your hands?"
"Do you think you're going to hell, Mr. Lyons?"
Rep. LaMonica McIver (D-N.J.): "Do you consider yourself a religious man?"
Lyons: "Yes ma'am."
McIver: "How do you think Judgement Day will work for you with so much blood on your hands?"
"Do you think you're going to hell, Mr. Lyons?"
February 10, 2026 at 5:31 PM
HOOOOOO boy.
Rep. LaMonica McIver (D-N.J.): "Do you consider yourself a religious man?"
Lyons: "Yes ma'am."
McIver: "How do you think Judgement Day will work for you with so much blood on your hands?"
"Do you think you're going to hell, Mr. Lyons?"
Rep. LaMonica McIver (D-N.J.): "Do you consider yourself a religious man?"
Lyons: "Yes ma'am."
McIver: "How do you think Judgement Day will work for you with so much blood on your hands?"
"Do you think you're going to hell, Mr. Lyons?"
Reposted by Benjamin Harnett
Washington Post trying to figure out if images can represent political ideas
February 10, 2026 at 2:13 PM
Washington Post trying to figure out if images can represent political ideas
Reposted by Benjamin Harnett
If you report on tech or are a tech-related decision-maker, please, for the love of Eris, understand that Silicon Valley runs on a hype cycle. Assume *everything* is bullshit until proven otherwise.
This would have been a good point for someone to have asked, is that really how things work? Or is vacuum in fact an excellent insulator? www.ft.com/content/a5cf...
February 10, 2026 at 1:44 PM
If you report on tech or are a tech-related decision-maker, please, for the love of Eris, understand that Silicon Valley runs on a hype cycle. Assume *everything* is bullshit until proven otherwise.
Reposted by Benjamin Harnett
No AI-involved system will work in favor of humans.
February 10, 2026 at 1:32 PM
No AI-involved system will work in favor of humans.
Are you really doing serious tasks with a black box algorithm that is demonstrably racist, sexist, filled with outdated information, trained by psychopaths, and completely hallucinates almost a third of the time and then say it doesn’t poison your work?
I think a subtle reframing might help people understand what it means to use LLMs for a task. Instead of “I had Claude scan and organize a folder of documents by keyword” say “I asked Hitler to scan and organize a folder of documents by keyword.” Not feeling so good about it now, are you?
February 10, 2026 at 1:39 PM
Are you really doing serious tasks with a black box algorithm that is demonstrably racist, sexist, filled with outdated information, trained by psychopaths, and completely hallucinates almost a third of the time and then say it doesn’t poison your work?
I think a subtle reframing might help people understand what it means to use LLMs for a task. Instead of “I had Claude scan and organize a folder of documents by keyword” say “I asked Hitler to scan and organize a folder of documents by keyword.” Not feeling so good about it now, are you?
February 10, 2026 at 1:33 PM
I think a subtle reframing might help people understand what it means to use LLMs for a task. Instead of “I had Claude scan and organize a folder of documents by keyword” say “I asked Hitler to scan and organize a folder of documents by keyword.” Not feeling so good about it now, are you?