"AI usage" Topic

Do you use an AI daily? weekly? monthly? never?

GROK gives this guess as to who is in each group: Power users vs dabblers. Daily use is usually occupational: code, drafts, research, customers, spreadsheets. Those people have a working model of what the tool does and does not do. Weekly/monthly use is more "I tried it," homework, one-off search, or colliding with AI slop in the wild. Rainey itself finds 70% of voters are annoyed to learn content was AI-generated and 41% say nothing about AI excites them. The people who live in the product see leverage; the people who trip over it see garbage, job threat, and data-center electric bills. That predicts a U-shape, which is what you have: daily least hostile, monthly most hostile, never somewhere in between.

I use it daily for work. It's not perfect, but workable at generating code.

However, I am bombarded daily with AI slop of the worst kind – its campaign season, so the daily mail is full of fliers that reek of AI – Candidates in odd poses, basically photo-shopped pictures, etc. I asked one the candidates who was obviously photo-shopped what he intended to do. His response was it wasn't worth the fight.

I fear that as AI gets better, being able to tell the fiction from the truth will get much more difficult. And how does one defend oneself if you are the target of the slop?

Daily – a mix of business and personal.

I would suggest that you define what "use" means. Are you talking about the AI generated results that you get from a general search on the internet or do you mean use AI to write something or to design something, etc. I think that they are two distinct uses. Just a suggestion about your suggestion. Otherwise, anyone who conducts a simple search for cuff type on Imperial Guard Marine uniform will have used AI, however, it is not really an intended use.

RJ, I a finding that GROK is fully conversant with gaming and with military history and uniform details etc. It doesn't judge and seems to treat every trivial inquiry I make seriously. Which of course it does, it is a machine.

But on other other hand, I haven't tried to use an AI assistant, but by this point both Chat and Grok know enough about me that if I asked them to recommend articles to read that they think I might find of interest, they'd do a pretty good job of it. And if I then gave feedback to refine their recommendations, pretty quickly they'd get very good indeed.

skedaddle, yes, but otoh my AI can also get better at screening for me. I think we will become as sophisticated about digital slop as we have mostly become about advertising. Like porn hard to define but I know it when I see it.

MY take ….

link

WF, yes, thanks, I have not automated like that but probably need to and perhaps will.

If I do a search, Yahoo gives me an AI answer, and these days many of the links are AI products. (Thank Heaven Yahoo stopped giving me incredibly inaccurate AI summaries of my e-mail!) I don't know how much I'd have to spend to get something "personalized" and evidently no amount of money will keep the current AIs from inventing answers and citing non-existent sources, so they're no real use to me for research.

I'd still love to see them generate decent fiction, possibly under human direction. And I'd pay semi-serious money to see them make substitutions in old movies. If they can't put William Powell and Eugene Paullette in all the old Philo Vance movies, couldn't they at least replace the tanks in "Patton" and "The Battle of the Bulge" with period-accurate models?

I tried using it for work. It's rubbish. Couldn't do a simple task.

Major b, I am curious when you tried it? The thing that is astounding is how fast it evolves. It was garbage 18 months ago and went to tolerable just over a year ago to ‘holy crap this is getting interesting' just this past March. AI is exponential in how it evolves and it will get scary before you know it.

I used more when I was working, but now in retirement I use it to mostly help me compile complex data equations and create images. I never use it to write as that has always been my passion and I'd rather write something than let an AI do it for me.

Major b, I am curious when you tried it?

A few weeks ago.

I tried it once at a party, but didn't really get the appeal so I never used it again.

I occasionally use it as a follow up to AI summarised search results, but otherwise never knowingly.

My ex work colleagues (IT) apparently use at least three different AI providers all at once, and it has become pretty fundamental to their jobs extremely rapidly.

My younger daughter is an oncologist and uses it all the time. Seems to have a lot of medical applications.

Martin, yes, we are going to be healthier, and pretty fast.

When I first showed my mother -- who was extremely bright, top of her class of 500 in a Houston high school in 1940 -- a computer, she asked well, what does it DO? I tried to explain that it doesn't do ANYTHING, but it helps with doing almost everything. What finally got her to see was likening it to a sheet of paper. Which also does not do anything, but try to get something done without it!

I think much of the current animosity to AI is a failure of imagination. AI just sits there waiting for you to ask it do do something. But pretty quickly a lot of us are finding new jobs for it to do.

(Buy of course there is the Skynet hyper-fearful imagination . . .)

Never that I know of.

I have some concerns about AI. But when I see those who are leading the opposition against it in the U.S., know how china is using money, Facebook and other applications to try to influence that negative view and I remember that EVEN if we quit, China, Russia, and other enemies of the United States, will continue to advance their AI development…. We'll… My vote is to continue.

You know, roughly a third of my "to do" list is scheduling appointments and reservations. I keep shoving them off because virtually all the bureaucracies I have to deal with shunt me to a computer-generated voice which doesn't "hear" well, offers me a menu of things I don't want instead of what I've requested, and is relentless in pushing "opportunities" I've already declined. Projects which would have taken under five minutes with a person now take half an hour if they can be accomplished at all. I have to fight through three layers of them, each with its own threats and advertising just to get my meds from the VA. They're not getting better: not in the last few months, not over the past year and not over the past 18 months. They're just getting more common.

Research? Doc, you've admitted even you wouldn't use them if the result mattered, because the tendency to provide non-existent information "sourced" to invented documents seems to be baked in. When I don't care about reality, I can invent my own facts, thank you.

Where I continue to have hope is genre fiction. I see no reason a journeyman writer couldn't soon be able to have the AI suggest a plot outline in a given writer's style, make changes when necessary, have the AI produce a Burroughs-style planetary romance, check to make sure it's adhered to the plot and sell me a decent piece of sword and planet for $200. USD (I'm not going to get a new one any other way.) So instead, we use the clanks for "research" and cast any novelist who is even suspected of using one into the outer darkness. It's insane.

One hedge. The private money-handling firms--banks, investment companies, credit card people--have systems which work at the level of a bright high school graduate, which is all the position requires. This may be why the billonaires imagine the things work: for them, they do. But that's not us. We're going to be dealing with the ones at the VA, the DMV, the airline reservations for unspecial people, the city "help" desk…Those are never going to get better because they're already just good enough to justify firing the humans, and that's all anyone ever cared about. Prompt and efficient service was never on the list.

They'll get faster. The machines will require more processing speed and more power. No one's even pretending they're going to improve my life--quite the reverse. I had to buy a new house this year: every communication device I own fill up with computer-generated confidence schemes. No telling when it will stop--if ever.

Tulip bulbs for the digital age.

EDIT: That was harsh--I don't think necessarily inaccurate, but I wouldn't say anything that bad about a person without hedging. I've been being put through the wringer of computerized bureaucracies for the past three months, and where Doc sees a coming golden age, all I feel is that I was beaten on by faster, harder sticks. Still four more on the current list, and I'm afraid it shows.

The part that consistently puts me on the opposite side from Doc in this is that so nearly as I can tell, everything in this life is a trade-off. Sometimes it's a really great trade-off, but you always pay something. When I hear Doc describing his new golden age--Doc, you never discuss risks, and you never mention the price. The computers will grow more powerful, and suddenly people and bureaucracies who never cared about the length or quality of my life will use them to help me live a longer better one. Why? My sources in medical research tell me that in order to get a grant approved today, they actually have to swear that it will help me less.

Talk to me about the trade-offs and risks, and I won't spend so much time checking the teeth of the alleged gift horse.

Robert, no no, I'd use them DIFFERENTLY if the results mattered. That is one of those trade-offs.

I am almost 80 and see multiple doctors regularly for several chronic conditions. Like most old folks I hurt all the time. And I understand frustration with medicos and to some extent share it. But I think AI is more likely to provide solutions to those problems than to add to them.

AI is an incredibly powerful tool and morally neutral as are all tools. So of course there are risks and trade-offs. But one risk is that fear -- stirred up by our enemies especially China -- will prevent us from using it to its full good effect.

Depends on what you mean by AI. Artifical Intelligence is a crap term. "I used AI" is about as specific as "I played a wargame". Was that Risk! or Pacific Sentry?

AI was …
10 years ago – Computer vision/object recognition
20 years ago – Netflix (and other) recommendations based on history of use
30 years ago – Deep Blue (almost*) beat Gary Kasparov
40 years ago – Grammar checkers
50 years ago – MYCIN/XCON medical diagnostic assistants
60 years ago – Chatbots (no, really)
70 years ago – AI term coined, created proofs of Principia Mathematica theorums
80 years ago – Alan Turing, Intelligent Machinery

* – If you want to talk about that, I teach a class on it

I accept the correction on research. But you're still being coy. Benefits are specific; "risks" are abstract, and real trade-offs aren't even mentioned. Handling the research differently is just technique: disappearing jobs and the relentless hounding only a machine is capable of are trade-offs, and only the ones we've discovered so far. (From my point of view, there are no benefits so far--though I expect there would be, if I lived long enough.)

As for meds, you're dreaming. If you want real improvement, work on the vetting process and the benefits--currently none--for discovering that a cheap unpatented substance will solve a medical problem. The people currently filling out grant applications have to swear the results will disproportionately benefit whoever's on the "oppressed" list this week. Being beneficial to everyone isn't good enough. So anything which actually benefitted either of us falls in the category of "happy accident." AI, like goodness, has nothing to do with it.

Major B has made many negative remarks about AI on TMP and I wonder where else. I suspect that no matter how hard he tries AI is going to mess and frustrate him on purpose. You are being punked.

Major B, an apology to the AI gods in order if you ever want to use it or maybe sacrifice a digital phone.

Wolfhag

This Seems like a good use to me:

The White House task force announced it is "using artificial intelligence and other technology to help detect fraud

Subject: JD Vance's Anti-Fraud Task Force to Boot 760K From Obamacare


link

I use some simple bits as required or if really bored. It's a reasonable search tool, but I'm not interested enough in other possible uses to get down into the guts, muscles and bones of AI. Most of my working life was playing with tech- testing, comparing, repairing, using, programming, coding and writing manuals for equipment, so I'm well over getting excited by shiny new tech toys. AI is there and I'll use it when I want to, but it won't necessarily replace other tools I use.

robert, I expect within a year -- and for all I know it is here now for some -- I will be "plugged in" to a medical service that monitors bp and glucose etc. in real time. If I choose to be. I already do a daily blood pressure reading that gets sent to a nurse, who calls me every month to discuss where I am -- and oftener if needed. That sort of screening will be, I am told, a game changer for many chronic conditions, and as easy as wearing a watch, and not expensive. I have chronic kidney disease and right now find out every three months how I am doing: that will become daily, constant. That will be available to millions and will save lives. And not just for geezers, but for athletes training, and for soldiers. We will KNOW a lot more than we did. I believe that expanding knowledge is a good thing -- although of course we humans misuse all of our tools and will continue to do so.

AI medical benefits from 20-30 years ago. These are not theoretical, abstract concepts. These are real operational cases, FDA approved use improving medical treatment for millions of people.

* In 1998, the FDA approved the ImageChecker system, making it the first commercial Computer-Aided Detection (CAD) system for screening mammography.
* X-Rays and MRI Analysis: Early Artificial Neural Networks (ANNs) were trained to assist radiologists in spotting lung nodules on chest X-rays or detecting brain lesions in MRI scans.
* The algorithms used in electrocardiogram (ECG/EKG) machines improved significantly between 20 and 30 years ago.Systems used mathematical feature extraction and neural networks to automatically identify arrhythmias, heart blockages, or signs of an active heart attack directly from the paper trace output, giving emergency room physicians immediate preliminary reads
* Clinical Decision Support Systems (CDSS)While 1980s systems relied on strict human-written IF-THEN rules (like MYCIN), the late 1990s and mid-2000s saw decision support transition into probabilistic and statistical models.
* DXplain: Developed at Massachusetts General Hospital, DXplain grew into a widely used decision-support tool during the 1990s and 2000s
* In the 1990s systems were integrated into hospital infrastructure to check for adverse drug interactions, flag incorrect medication dosages, and suggest antibiotic regimens based on institutional resistance patterns.
* Machine learning algorithms (such as Hidden Markov Models and Support Vector Machines) became critical tools for searching massive DNA sequences, predicting gene structures, identifying protein-coding regions, and mapping genetic mutations associated with diseases like cancer.
* Surgical Robotics and Motion FilteringIn 2000, the da Vinci Surgical System received FDA clearance for laparoscopic procedures. Human surgeons controlled the robotic arms, the system relied heavily on real-time computational algorithms to eliminate natural hand tremors, smooth out physical motions, and provide scaled micro-movements inside the patient's body.
* Researchers began implementing early machine-learning models trained on bedside vital signs (heart rate, blood pressure, oxygen levels) to predict clinical decline or septic shock hours before visible symptoms appeared, laying the groundwork for ICU monitoring.

My son is an Army Combat Medic. Here's what he is trained to use today:

* Algorithmic Triage: Predictive models assess patient vital signs, injury severity, and hemorrhage risk to prioritize casualties during mass casualty (MASCAL) events when resources are overwhelmed.
* Field-deployable ultrasound devices (POCUS) use computer vision to instantly guide frontline medics with limited training to spot internal bleeding, collapsed lungs (pneumothorax), or shrapnel without needing a radiologist.
* En-Route Care Support: AI alerts transport crews to early signs of septic shock, airway compromise, or internal bleeding long before physical symptoms appear.
* Military health agencies deploy hands-free AI tools that record verbal medic-patient interactions, structure notes, and automatically draft field care records (like the Tactical Combat Casualty Care TCCC card). This frees combat medics to focus their hands and eyes on saving lives.
* VA System Integration: Across domestic military hospitals, ambient AI scribes automatically convert doctor-patient interactions into EHR entries to reduce administrative burnout.
* Predictive supply-chain algorithms anticipate local demands for blood plasma, surgical kits, and oxygen based on combat intensity, ensuring field hospitals do not run out of critical supplies.
* Smart wearables track a service member's biometrics (heart rate variability, core temperature, sleep, fatigue) to predict heat illness, exertion limits, or early infection before a soldier drops during operations.
* AI models analyze speech patterns, facial expressions, and vocal inflections to screen returning troops for early signs of Post-Traumatic Stress Disorder (PTSD), traumatic brain injury (TBI), or depression.

Note that all these are tools that medical professionals use. They don't replace medical professionals; they augment their ability and scope.

eto, yes, great stuff

Eto, no one who believes the VA practices any sort of "systems integration" has ever visited a VA facility. I spent part of September being poked and prodded by the Lincoln VA--including a fresh blood draw--because they evidently couldn't access the exact same procedures done in August in Fort Wayne. It's not uncommon to be force-fed the same "lifestyle questionaire" three times in a morning because no one knows what the previous station has done.

That's the part I have personal knowledge of. The rest is second-hand, or is going to happen some day, you bet. I can remember when electricity was going to be too cheap to use meters.

Well, I go to the VA and I used to audit their IT systems. The transcription systems are there and working. Have been for quite a while.

But, if one person who didn't know what they were looking for didn't see them, then obviously, they don't exist.

Also, read the post. It doesn't say every system is integrated. Your example has nothing to do with AI, it is simply data sharing.

The post says that AI transcription services work in the background and transfer information into record systems so doctors don't have to take notes then type them up.

So, I guess your post is "a bad thing happened to me at the VA once, so everyone there is incompetent and advanced technologies don't exist".

Yes, Eto. And your post is "the VA's check cleared, so surely things work--and if they don't, it must be the fault of the non-tech people, because it can't possibly be the fault of the machines or the techies."

Yes, Eto. And your post is "the VA's check cleared, so surely things work--and if they don't, it must be the fault of the non-tech people, because it can't possibly be the fault of the machines or the techies."

Please point out the part where I said everything at the VA is good. Please point out the part where I said something was somebody's (anybody's) fault.

Just because you want to overgeneralize from a small sample doesn't mean you can accuse other people of doing it.

I have, in fact, no idea whether the AI can read doctor's hand-written notes with sufficient accuracy. Do you, really? Watching what OCR does with typefaces does not fill me with confidence. Get back to me about the AI allocating medical supplies after the next conventional war.

What I have noticed is that, if the AI is as capable as its advocates insist, it's right up there with fire and the wheel--well ahead of nuclear power. But as soon as anyone asks whether the benefits outweigh the risks, or whether there might be other ways to, say, improve medical outcomes, we're dismissed as troglodites.

Who, in the end, owns and runs the AI? (Mumbled answer): "Politicians, left-coast techies and a few tech billionaires."

Haven't these people already sold us a few glorious futures which turned out to work better for them than for us?

And if the AIs turn out to be autonomous and not just scapegoats, who tught them ethics? (Mumbled answer): "Politicians, left-coast techies and a few tech billionaires."

Any of those you'd trust with a child's piggie bank? I thought not.

A glorious future may yet await, despite the rather disreputable prophets. But proceed with caution, keep power dispersed, and don't expect the machines to solve messes which are at their root poor organization and lousy incentives.

robert, per your last, yes, humans are still fallen and flawed, the root of most of our disorders.

I have, in fact, no idea whether the AI can read doctor's hand-written notes with sufficient accuracy.

Not sure about AI reading doctor's handwritten notes.

Very sure about you not being able to read a post on this thread. Nobody mentioned that before your post that this is from.

And, of course, you refuse to answer simple questions about where the things you are refuting were posted in the first place. And went of on a rant about something else different.

Who, in the end, owns and runs the AI?

Who owns the software? The same person who owns any software. The people who built it.

Who runs it? The same people who run any software. The people running it.

But to address your strawman, I guess you mean billionaires, left-coasters, and politicians like …

* Anthropic (Claude): Founded by former OpenAI researchers as a public-benefit corporation. While Amazon and Google are minor investors, Anthropic operates independently and focuses heavily on AI safety, alignment, and long-context capabilities.
* Mistral AI: Based in France, Mistral is Europe's leading independent frontier lab. They build open-weight models (like Mixtral and Codestral) designed to run on private infrastructure.
* Cohere: A Canadian enterprise-focused AI platform specializing in retrieve-augmented generation (RAG), search, and multi-lingual language models for business applications without lock-in.
* Hugging Face: Often called the "GitHub of AI," Hugging Face is an independent hub hosting hundreds of thousands of open-source models, datasets, and apps created by developers worldwide.
* Ollama: A completely free, open-source platform that allows individuals and enterprises to run AI models (like Llama, Mistral, and DeepSeek) locally on personal computers or private servers without sending data to external APIs.
* Groq: An independent hardware and cloud platform running custom "LPU" (Language Processing Unit) chips designed specifically for ultra-fast, low-latency AI inference.
* Together AI & Replicate: Independent developer platforms that host open-source AI models as simple APIs, allowing creators to build products on open weights without renting raw servers.
* Perplexity AI: An independent answer engine and research assistant that routes queries across various underlying models to replace traditional search engines.
* Midjourney: An independent, self-funded AI image generation tool run by a small core team, famous for operating without major corporate or venture capital control.

And if the AIs turn out to be autonomous

This is just so wrong, it can't be responded to in a reasonable manner.

So … What does anything in any of your posts have to do with any of the actual content of the posts you are responding to?

Possible I misunderstood "ambient AI scribes automatically convert doctor-patient interactions into EHR entries." Are they working from voices? (And if it's neither written nor verbal, just what sort of "doctor-patient interactions" are we talking about?)

I will stand by my contention that while I may sometimes pay for software, I am not the person the software works for. In any conflict of interest, the advantage lies with the person writing the code and his employer, not the poor devil hoping for unbiased research, or even for the AI to tell him a story. If my car gets lousy mileage or breaks down frequently, at least I know. If my search engine is holding out on me or subtly altering search terms and tilting statistics, how do I even find out?

As for the players, I gave three criteria. How many of the organizations you list are using programmers from the same cultural pool which gave us images of black females signing the Declaration of Independence and Asian SS? And how would YOU measure relative importance? Number of programmer-hours? Investment in dollars? Something else altogether? Would as much as 20% of the AI effort not fall into my three categories? (And how do I ensure they're the ones conducting my searches?)

For the record, this was not my feeling even 20-25 years ago. Now, I'm feeling more and more like Treebeard--"not on anyone's side, because no one is on my side." It's been a rough quarter-century, and in terms of using computers to make my life easier high tide may have 10-15 years ago.

Are they working from voices?

That's the ambient part. Written notes are not ambient.

If my search engine is holding out on me or subtly altering search terms and tilting statistics, how do I even find out?

Check the references the way you are supposed to, no matter where the info comes from?

How many of the organizations you list are using programmers from the same cultural pool which gave us images of black females signing the Declaration of Independence and Asian SS?

That was Google Gemini, so … none.

And how would YOU measure relative importance?

No idea what this is asking.

Would as much as 20% of the AI effort not fall into my three categories?

Go back to my first post here, where I point out "AI" is a useless term when used to indicate "every AI application". AI has been around for a little more than a century, and encompasses a lot of things.

Your spellchecker is likely not a static pattern matcher, but most likely a probabilistic context matcher. That's AI.

And how do I ensure they're the ones conducting my searches?

Don't just click "Accept"?

Eto, I really hope you're right on every point. Sadly, nothing you've posted lessens my fear of a future in which virtually all broad research filters through software created by a social/political monoculture. Books and articles which dispute the core points of the monocultural consensus will be determined to be unreliable for just that reason: they'll be disregarded in summaries and not listed in searches for material on a given subject. Libraries will remove such work to storage sites you can't physically reach, then remove them altogether since no one requests them. That's if they ever make it to the shelves. Most library holdings are shipped to them from a dwindling number of picker firms. It's another easy way to maintain a consensus. And while librarians complain a lot about censors, many of them are quite good at making sure only books they agree with are purchased and prominently displayed.

Writers complain bitterly that five major publishers are too few. In twenty years, will there be as many as five widely-used search engines? And will any of them play the game straight?

We have the technical capacity to create a world with almost unlimited access to ideas and research. But if we gum up the software, we're back to samizdat.

You can't debate fear, so that is a closed door.

None of my statements were predictions for the future.

We already have biased selectors feeding recommendations. Been like that a little over 30 years. They work because that is what people want.

Pretrained Transformers (the PT in ChatGPT) are all biased. You want bias. Bias is how you discriminate among options. The question is what is the bias and how does it align with your use case.

A PT application heavily biased toward only rigorously demonstrated scientific knowledge would be a poor tool to help write a science fiction novel or wargame scenario. Or to search for existing scifi … novels, movies, TV series.

A prediction for you: A monocultural control won't happen. Adversarial political power bases need adversaries. Some movements have had limited success shifting them and inventing them. But if you don't have solutions, you need to blame your failure on someone.

The bulk ubiquitous ingest with current PT applications doesn't really imbue a specific cultural or poltitical bias. They end up biased toward view with the largest number of whiny little bitches. They flood the input signal.

This is not a new problem in social information management. "Happy people don't complain. Grumpy people don't shut up."

Robert,
I'd still love to see them generate decent fiction, possibly under human direction.

There is a way. Put this into an AI prompt:
i want to use AI to write science fiction taking place in the far future, not the past. However, because of the bias and data not being verified because it is fiction and not true it needs human creativity. Here is what I'd like to do: I want to be able to access the best in science fiction in the last 100 years including books, movies, articles and podcasts and put them in a remote or cloud repository offline out of the internet. Then, with the help of AI under my control, use it to get ideas for plots, situations, ideas and human and non-human interaction. Please give me direction and detailed instruction for each step and the apps to use as I'm a real novice with AI and the various tool to create a searchable repository of information without having to worry about using inaccurate references and hallucination. do you understand?

I'm sure someone can come up with a better one.

Wolfhag