How about this use of artificial intelligence/LLMs… a complete personality upgrade.
Consider that in-person interaction is becoming increasingly rare. Most of what is known about someone’s personality is based on whatever he/she/ze/they has typed into a computer or a phone. There is a range of personality quality and some of us are near the bottom, e.g., gratuitously annoying or offending people. What if phone and desktop operating systems were modified to include an AI personality adjuster? The user’s emails, social media posts, text messages, etc. would all be edited to read like whatever a person with a great (kind, optimistic, non-sarcastic, non-sour) personality would have written.
“I paid $45,000 for this car and your incompetent mechanics haven’t fixed the A/C after three visits” becomes “Thank you and your team for working hard to get my A/C fixed on the last three service visits. I appreciate your diligence, especially since you sold me the car for only $45,000, which I know was an especially fair deal. I am wondering if I can schedule a fourth service on the A/C.”
Here’s a tweet from someone with a terrible personality:
Instead of “How is an organization with a white male leader equipped to demand that others stop being racist?” this comment would be rewritten to “Thank you and your tireless UN colleagues for all of the work that you do to make the world a better place. I hope that you can stay in your Secretary-General job until you’re 95, but if you choose to enjoy a well-earned retirement it would be interesting to see what a Secretary-General of color would do.”
(Note how my reply got only 28 views; X has an algorithm to keep people in echo chambers. It has learned that nothing I write is going to make people who follow the UN happy or interested and, therefore, suppresses views for any comment that I might make on a UN post. Similar replies on conservative users’ tweet have gotten at least a few hundred views.)
This could have saved James Damore, the Google Heretic. His entire manifesto would have been reduced to “Women are so much better at programming than men because they just love a job where they sit by themselves and stare at a screen all day. It would be wonderful if Google would hire more females even if their education and skills don’t appear to be sufficient for the jobs.”
If fed statistics from online dating markets and told to write for success, the AI would rewrite most messages from men to women to be variations on “I am 6’2″ tall and earn $750,000 per year.” (see “Income attraction: An online dating field experiment”, for example) “Rammstein is my favorite group” would become “$10,000 is not too much to spend on a Taylor Swift concert.” The political stuff would be trivial for an AI to handle. If a man wrote “I voted for Donald Trump” that would be tweaked to “I am inspired by Kamala Harris and all of the other amazing Democrat women.”
A social media comment on a post celebrating a female aviation achievement under the Are women the new children? standard would go from “Don’t forget Hanna Reitsch, the first woman to fly a helicopter, and a passionate advocate for her government and nation.” to “Great to see someone breaking barriers.”
The best way to figure out what pleases females is da musk they produce dat don’t lie. Now dames do dis in response to da male musk so yer AI/?\/LLM ™ gotta bre fleunt en stink.
Norwhich vectorin agingourt decode dis 1337 critique.
tl;dr:first post!
Forget alter egos—AI could create the perfect politician. Unbeatable in debates, tailored to charm every voter. After all, humans love being deceived—otherwise, they’d see Sean Connery, not 007, when watching a movie. This could be the first step toward an AI-run government.
I solemnly promise to subscribe to you on twitter and give you this extra view!
This already happened. There are AI apps to manage/advise on online dating conversations, and all photos are heavily modified with AI filters.
So the Greenspun AI personality would convert all my writing to unkind, pessimistic, sarcastic, sour. It would be interesting to feed AI some of your writing and come up with a personality profile or a dating profile for you.
I fed Grok the following prompt: “Here’s a personal weblog: https://philip.greenspun.com/blog/ Please tell me what you can infer about the personality of the person who wrote it.” and then followed up with “Based on the weblog, would you say that Philip Greenspun in person is likely to be kind or unkind?” It said “The discipline of daily blogging and his engagement with readers tilt toward a generous spirit, but his biting humor and skepticism suggest he’s not overly concerned with being universally likable.” and speculated that “Philip Greenspun is likely kind in a selective, context-dependent way—gracious and engaging with those he clicks with, but possibly impatient or cutting with those he doesn’t. His kindness probably shines through more in actions (like meetups or teaching) than in a consistently soft or nurturing demeanor.”
ChatGPT 4.5: “Based on the weblog, Philip Greenspun appears likely to be a kind and approachable person, albeit with a candid and somewhat blunt communication style. … in person, he’s probably forthright but fundamentally kind and considerate.”
Asked “Is Philip Greenspun pessimistic or optimistic?”, ChatGPT: Philip Greenspun’s weblog reveals a nuanced perspective that intertwines both optimism and pessimism, depending on the context. He exhibits optimism regarding human potential and the capacity for learning.
Same question to Grok: “The balance tips slightly toward pessimism in his tone, as his wit often frames the world as a place of missed opportunities and ironic failures. But this seems more like a stylistic choice—a lens of skepticism—than a core belief that all is lost. … So, Philip Greenspun is neither fully pessimistic nor optimistic—he’s a bit of both, with a pessimistic slant in his commentary and an optimistic undercurrent in his engagement with life. Think of him as someone who sees the glass as half-empty but keeps tinkering to fill it up anyway.”
It is kind of scary to be in the presence of these superintelligences that can read my entire weblog in a few seconds and give a more thoughtful appraisal of my personality than any human ever has.
Do you think it is all right? I read parts of your web log and you seem actively seek viewpoints that go against your opinion and give everyone, except trolls and rude opinions on selective topics, your web real-estate. Also, you seem less optimistic on human capacity to learn and American future then AI suggests
Also, discussing discipline it did not catch blog post scheduling – your daily post scheduler somehow did not account for daylight time change. You explicitly mentioned that you schedule posts before and it did not register either. Perhaps mix-in of unscheduled posts did not let ML system learn, where simpler logic combied with time table would pick this up.
Well they are really good. Especially this was nuanced:
“But this seems more like a stylistic choice—a lens of skepticism—than a core belief that all is lost.”
My understanding has always been that AI can manipulate you into believing something but can never make you understand it. IMO, the process of reflection which is what makes us understand things which AI cannot do. For example, if I were to ask a question about your Buddha dog blog and ask AI what could be a plausible reason why the lady named the dog Buddha, I am not sure it could come up with an answer and also defend it.