If you'd like an essay-formatted version of this post to read or share, here's a link to it on pluralistic.net, my surveillance-free, ad-free, tracker-free blog:
As far as I can tell, this dialog between MacArthur prize-winning mathematician Terrence Tao and Chatgpt about "the Jacobian conjecture counterexample" is very impressive:
Now, the clause "as far as I can tell" is doing a lot of work in that sentence. I am reasonably math literate, up to first-year calculus and a lifetime spent around my father (a mathematician). However, I have never heard of "the Jacobian conjecture," and while I know what all the words in the first paragraph of the relevant Wikipedia entry mean, I can't parse any of the sentences they form:
https://en.wikipedia.org/wiki/Jacobian_conjecture
In other words, I lack the discernment to evaluate the output of the chatbot that Tao exchanged theories with. If you showed me an equally opaque transcript of a "conversation" between a chatbot and a crank with AI psychosis whose math made no sense whatsoever, I couldn't make an a priori judgment about which one was a solid piece of mathematical theorizing and which one was a math-flavored word-salad.
As many skilled programmers can attest, chatbots can produce very useful output – but as even the most ardent AI-assisted coder will admit, chatbot-written code is also full of baffling, obvious errors (and subtle, hard-to-spot ones):
These errors (which the industry wants us to refer to as "hallucination" – a whimsical, obscuring, anthropomorphizing euphemism) are the reason that reliable AI use requires the discernment that comes from skill and expertise. I use a local chatbot to spellcheck these posts. Chatbots spot all kinds of typos that regular spellcheckers miss:
There's a reactionary group of strangers who seek me out to tell me that I'm a bad person for doing this. These are pointless conversations, mostly because I can barely make out a word over the scraping sounds of all the goalpost-moving these scolding strangers engage in.
They start by insisting that I'm burning down the planet by running a low-CPU load piece of software on my own computer. After I explain that running a chatbot on my machine uses no more carbon than, say, applying a blur effect to an image in my image editor, they tell me I'm unwisely giving my private data to the AI companies. Then I show them the network logs that demonstrate that my local chatbot doesn't send or receive any network data.
Then they turn to the supposed cognitive effects of using a chatbot to find typos in an essay. I explain that I'm not asking an AI to write things for me or explain them to me – I'm asking it to point out where I've forgotten to put a period at the end of a paragraph, or fatfingered a word like "ever" as "every." I even send them the "before" and "after" of an essay after I've corrected some chatbot-identified typos in it:
https://craphound.com/before.txt
https://craphound.com/after.txt
This is when things get increasingly pointless. My interlocutors come up with farcical reasons why it's immoral or dangerous to use this LLM-based spellchecker. They say I'm using too much compute and that I could use a simpler piece of software to do the same thing (which is both untrue and silly – I also run a journaling filesystem on my computer that is vastly overpowered for editing a textfile – who cares?). Or they insist that the mere act of making copies of published works in order to count their elements and the relationships between them is a sin, despite the fact that this standard would kill search engines, the Internet Archive, and the Oxford English Dictionary:
I mean, by all means let's hate the AI companies and work to end their disgusting campaign to pauperize creative workers, but let's not fall into the trap of siding with the media bosses who insist that the salvation of creative labor will arrive when Sam Altman pays David Zaslav for the right to cram the entire Warner catalog into Openai's chatbots:
Above all, my interlocutors continue to insist that my LLM-powered, local, open source chatbot spellchecker will make me a worse writer. It's a very strange insistence. My first word processor was a program listing published in a magazine I bought at a corner store and laboriously typed into my Apple ][+. In the 40+ years since, word processors have gotten lots of new features, many of which I thought were useful and many more that I found annoying. There were even some of these features that made the writers who used them worse at writing, in my (expert) judgment.
But from the very start, I knew that you couldn't just trust a spellchecker to correct your documents. I mean, I'm a science fiction writer. I started making up silly words decades before coining "enshittification." I've been telling spellcheckers to fuck off since I learned to type. If you aren't a good writer, spellcheckers are dangerous, and the more "advanced" the spellchecker is, the more dangerous it is.
A few of my collaborators insist that I use Office 365's AI-enabled version of Word to work on documents with them. It's maddening. I estimate the ratio of good suggestions to bad ones that M365 insists on shoving into my face at about 1:100. It's practically unusable – so much so that I often copy the block of text we're working on into a text editor, make my changes, then paste it back into the Word window.
If I were to accept even 10% of these suggestions, my work would be made significantly worse. Putting chatbots into Word pushed it from "annoying" into "enshittening." I certainly understand how relying on a chatbot to make edits to your work could make it worse.
That's where discernment comes in. I have written more than 30 books over the past 25 years. I have lots of experience defending my word choices, and not just against the mechanical judgments of a high-handed spellchecker, but also against overreaching copyeditors and paranoid publisher's lawyers. I know which words I want to write, and I know why I want to write them – and I know when a suggested fix is a good one and when it's wrong or stupid or just plain clunky. When it comes to writing, I have discernment.
That's not true when it comes to higher math. I would no more ask a chatbot to explain "the Jacobian conjecture counterexample" than I would tell my writing students to get a chatbot to suggest ways to fix their stories:
I don't know nearly enough about math to ask a chatbot to explain it, or check my work, or even assemble a bibliography of human-authored works I should work my way through if I want to learn about it. If I wanted to understand "the Jacobian conjecture counterexample," I would set aside several days and work my way through that gnarly Wikipedia entry and its references and blue links to related concepts. If I really wanted to understand it, I'd enroll in a course at the Open University or Khan Academy.
All of this has been obvious to me since I first encountered LLM-powered bots. If you understand a subject really well – well enough to discern useful bot output from defective bot output – then bots can be useful. Sometimes very useful, mostly ordinarily useful. For example, I've been writing Pluralistic for about 6.5 years now. I've written 1,683 posts now (1,684 after I hit publish on this one), and the corpus is now getting large enough that I sometimes struggle to find a post I'm trying to reference, even with all my careful tagging and my extensive knowledge of WordPress's URL-line options for searching the database with tag and keyword combos.
I've been toying with the idea of exporting my whole corpus and shoveling it into a local chatbot, so that I can type, "Which post did I talk about the evils of showing people your chatbot output in?" and get a link to the correct essay:
(Don't follow this link! I will be referencing the essay it goes to shortly; I struggled to find it when I sat down to write today; I'd accidentally tagged it with "at" instead of "ai" and missed the typo when I published it.)
There are very few subjects I have more discernment over than "essays I have written." If I ask a chatbot to tell me which post I'm thinking of, I will instantly know which of its guesses are correct and which ones aren't. No one in the universe is better qualified than me to perform this task. No one ever will be.
Now, as it happens, I know exactly how badly a chatbot can screw up when it comes to my own work, because strangers insist on asking chatbots about me and then, for reasons I find baffling, they send me the output. Please don't show anyone your chatbot transcripts unless they ask to see them. It's embarrassing at best and annoying at worst:
(There's that reference I promised. You can follow the link now!)
Again, discernment is everything when it comes to getting useful work out of a chatbot. If you don't know anything about my work and you ask a chatbot to explain it to you, you will likely be badly misled. If you are familiar with my work and you ask a chatbot for the best examples where I explain a given subject, you may get a good answer, and if you get a bad one, you'll know it.
The centrality of discernment to productive AI usage is obvious, and that's why I find the insistence that AI can be used as a teaching assistant (or worse, a teacher) so baffling. By definition, a student isn't an expert on the subject they're studying. That's the whole point of studying – to acquire knowledge and thus discernment. Asking students to learn via chatbot explanations is both incoherent and dangerous.
Doubtless, there are ways that teachers might find chatbots useful, but for Christ's sake, don't use them to teach. There's plenty of ways teachers can use chatbots without asking students to learn from them.
Here's an example. My daughter graduated from a big, typical American high school a couple years ago, and I spent her high-school years getting progressively angrier about the bad compromises that her teachers were forced into.
Between "Common Core" and "Advanced Placement," the US system has been highly standardized. Teachers are under enormous pressure to teach specific aspects of specific subjects in a specific order, and students are told that their future life chances turn on their ability to pass high-stakes tests:
This gives rise to many frustrations for teachers and students alike, but nothing got my dander up so much as my daughter's math teachers' testing practices. In all of my kid's higher math classes, teachers had a single, prized set of tests, and lived in fear of these escaping into the wild and turning into cheating aids. As a result, teachers collected students' math exams and quizzes and did not return them. Students sat exams, worked through the problems and got their grades – but were not allowed to take home their tests to see where they went wrong.
Look, I know I'm no mathematician, and I know I'm not a math teacher, but I know enough about pedagogy to know that this is crazy. This is like trying to get better at archery by loosing arrows at a target but not checking to see where they hit. It's bananas.
I also understand why the teachers felt they had to do it. Writing test questions that test for specific concepts in a specific order is a lot of work, and generating new tests for every class is the kind of task that would consume time better spent on lesson planning and meeting with students.
It's easy to imagine a teacher who creates prompts for each test question that cause a chatbot to emit a new test paper for each class, along with answer keys. These questions are easily validated by a skilled teacher, who definitionally has the discernment to know whether a test question fits the bill. I could even see vibe-coding a little app to spit these questions out – though again, I would want the teacher to work through the questions each time to make sure they were sound.
Both my parents are teachers. My brother is a teacher. I teach every now and again. Teachers do a lot of repetitive, unrewarding work. They also do a lot of difficult, creative, extremely important work. Good teachers have the discernment to sort good classroom materials from bad ones. They do that already, because just as you don't need an LLM to generate bad spellchecker suggestions, you also don't need an LLM to generate sub-par educational materials. There are plenty of "educational" publishers who'll do that all day long.
AI is a normal technology. That means there are times when it is useful and times when it is pointless or actively harmful. One rule of thumb for chatbots is that they can only provide useful information to experts who have the discernment to ignore the defective output that LLMs always emit. That means that the dream of chatbots as replacements for teachers is a nightmare.
Getting rid of teachers because we all have chatbots is like getting rid of doctors because we all have the plague.
AI can make our work faster and better, but new research shows it may be quietly weakening how well we think. Studies from Wharton, Carnegie Mellon, and Anthropic found that people who lean on AI too much stop learning, make more mistakes when the AI is wrong, and lose skills they never got the chance to build. The fix isn't avoiding AI — it's using it wisely: let it handle busywork, but keep the thinking, learning, and decision-making for yourself.
Why it matters: Good communication depends on real understanding, not borrowed answers — if we let AI think for us, we risk losing the judgment and self-knowledge that make our words and decisions trustworthy.
Read the full story to learn how to use AI as a thinking partner, not a replacement for your own mind — because technology only moves forward when humans stay in charge of it.
AI agents — systems that can plan, act, and make decisions on their own — are getting smarter and more common. But The Atlantic warns that the real crisis isn't about what AI might do wrong. It's about what we might stop doing ourselves. When machines handle more of our choices, we slowly give up the practice of thinking, deciding, and leading — and that loss is hard to notice until it's too late.
The fix isn't to slow down AI. It's to keep humans firmly in the driver's seat. Technology only moves forward when people stay engaged, ask hard questions, and hold the final call. Every tool, no matter how powerful, needs a human hand to guide it toward good ends.
Why it matters for communicators: The way we talk about AI right now sets the culture for how it gets used. If we frame AI as a replacement for human judgment, that's exactly what it becomes. The better story — and the truer one — is that AI works best as a partner, not a boss.
Want to think more clearly about who's really in control of the technology shaping your life? Read the full piece in The Atlantic and decide for yourself.
How AI Damages Work Relationships—and Where It Can Actually Help
AI can help at work, but it comes with a hidden cost — damaged relationships. A new HBR study found that when people use AI to handle workplace conversations, colleagues feel less trusted and less valued. The tools that save us time can quietly erode the human connection that makes teams work.
The good news: AI does not have to be a relationship killer. Used wisely — for routine tasks, scheduling, or drafting a first pass — it frees people up for the conversations that actually matter. The key is keeping humans in the loop, being transparent about when AI is involved, and never letting a machine handle the moments that call for empathy, judgment, or genuine care.
Why it matters for communicators: Every message you send either builds or breaks trust. When your audience can feel the difference between a human voice and an automated one, authenticity becomes your most powerful tool.
Ready to find the right balance between AI efficiency and human connection? Read the full article and see exactly where to draw the line →hbr
CEOs aren't seeing any AI productivity gains, yet some tech industry leaders are still convinced AI will destroy white collar work within two years
A sweeping survey of 6,000 executives across the US, UK, Germany, and Australia — conducted by the National Bureau of Economic Research (NBER) — found that 70% of companies now use AI, yet nearly 90% report seeing no measurable gain in productivity or employment over the past three years . Top executives are among the lightest adopters: a quarter of CEOs and CFOs don't use AI at all, and two-thirds log just 90 minutes of use per week at most — a striking disconnect between the boardroom hype and on-the-ground reality . The findings align with earlier MIT research showing 95% of AI pilots failed to deliver productivity gains, suggesting the gap isn't simply a matter of early adoption — it may reflect a deeper structural mismatch between what AI can do in controlled settings and how it performs inside complex, human-driven organizations .
The risks ahead, though, may outpace the gains. Microsoft AI CEO Mustafa Suleyman has predicted that AI will reach "human-level performance" on most professional tasks — legal, accounting, project management, marketing — within 12 to 18 months, with mass automation to follow . Senior executives project the technology could eliminate 1.75 million jobs over the next three years, even as their own staff expect AI to increase employment — a fundamental disagreement about the future that signals serious accountability gaps within organizations . Without deliberate governance, the same tool companies are deploying to boost output could quietly erode the human oversight that keeps decisions fair, contextual, and grounded in real consequences .
Why it matters (for communicators): These findings are a clear-eyed warning: AI without a human in the loop isn't a productivity strategy — it's a liability. Communicators must champion the human constraints — judgment, ethics, accountability, and context — that keep AI from automating its way past the guardrails that protect both people and organizations. The most dangerous version of AI isn't the one that underperforms; it's the one deployed at scale with no one meaningfully in charge of what it says, decides, or displaces.
Which is it? No productivity or white collar workforce Armageddon?
