Secret To AI Profitability Is Hiring a Lot More Doctorates (yahoo.com) 85
As tech giants struggle to profit from AI, a growing industry of specialized AI training firms is emerging by hiring doctors, radiologists and other experts to develop commercially viable applications. The $20 billion data services sector, projected to grow 20% annually, is attracting major investment by focusing on high-value, specialized AI applications.
Companies like iMerit and Centaur Labs are recruiting specialists worldwide, from radiologists in Kazakhstan to agricultural experts in Bhutan, paying premium rates for domain expertise rather than basic data processing. While Microsoft and Alphabet post losses on AI development, this specialized sector is finding profitability by bridging the gap between raw AI capabilities and practical business applications.
Companies like iMerit and Centaur Labs are recruiting specialists worldwide, from radiologists in Kazakhstan to agricultural experts in Bhutan, paying premium rates for domain expertise rather than basic data processing. While Microsoft and Alphabet post losses on AI development, this specialized sector is finding profitability by bridging the gap between raw AI capabilities and practical business applications.
Why can't they just ask... (Score:5, Insightful)
Meanwhile, in addition to the drain on electricity, infrastructure, & hardware resources, they're now proposing to be a drain on our best thinkers. Where's this "intelligence" thingumy they've all been working on? It doesn't seem to be working very well, does it?
Re: (Score:3, Interesting)
... because the ones who were dumb enough to fall for crypto have already tossed their remaining assets into the AI money pit.
Re:Why can't they just ask... (Score:4, Insightful)
Same reason why they can't ask doctor students to be more useful. They need to be trained first.
AI image recognition is already pretty good at basic medical stuff. You can try it with Grok already, it's now free for a few tries. Not "doctor" level good yet, but significantly better than "I looked it up on webMD and I can't afford to/don't live a place with access to a doctor over this".
Re: (Score:2)
Re: (Score:2)
Western system of medical services is nothing short of utterly unreachable for most of the world's population. And Grok doesn't discriminate based on where you live. As long as you can connect to the internet and have a phone to take pictures and send them to it, it will give you a basic diagnosis.
Re: (Score:2)
Of course, "access" in the USA is the leading cause of bankruptcy, which I don't think is an issue that AI can adequately address. Or perhaps you're suggesting that the USA needs more backstreet doctors? It certainly looks like backstreet abortions are going to become commonplace once more.
Re: (Score:2)
US has excellent medical care. I am in a nation with "good" medical care, and I've sat in queue for about six months at this point for a basic dentist appointment. Still in it, no idea when my time arrives, they said when I reserved my place in queue that waiting time is 5-7 months. Getting a basic doctor's appointment is a matter of fighting with nurses that your problem actually justifies doctor's time and a wait of at least a month. We're pretty good by socialized medicine standards, the likes of UK and
Re: (Score:2, Informative)
In the US, we also wait 6 months for a dentist. We wait a month for a basic doctor appointment. The stories about excellent health care in the US are outliers, or stories from the 1900s.
Re: (Score:2)
Internet exists, and I can fact check this. Let's see:
https://www.statista.com/stati... [statista.com]
So rounded to a day, 10 days for general dentistry, 8 days for a specialist dentistry.
So that's a "liar liar, pants on fire" from me doug.
Re: (Score:2)
> Average patient wait for select U.S. dental appointments in 2023, by type of dentist
Cool, in Canada the over half of Canadians wait 3 days or less for minor medical issues (like the ones a regular doctor can handle), which is less than 10 days and certainly less than the month you say it is.
Stats from the canadian government [statcan.gc.ca]
Join my book club.. doug?. [amazon.com]
Re: (Score:2)
Ah yes, government checked itself and found it's doing really well through careful crafting of terminology.
Meanwhile, reality:
https://www.statista.com/stati... [statista.com]
Would you like a fire hose to put out the fire engulfing those pants doug?
Re: (Score:2)
Who is doug?
Re: (Score:2)
This is always the scape goat isn't it? "Yes, we know our health care is one of the best if not the best in the world. But there are people who can't afford it in lower middle class, and they crash the average, and because of that everything sucks".
It's not like around 80% of the nation gets to enjoy healthcare quality I can only dream of. In a supposedly one of the best nations on the planet healthcare wise overall.
Re: (Score:2)
Tell us more about your lived experience.
Re: (Score:2)
So rounded to a day, 10 days for general dentistry, 8 days for a specialist dentistry.
It depends a lot on what specific service you need. Where I am, appointments for fillings and crowns can usually be booked within two or three days. Dental cleanings are done by separate dental hygienists, and have to be booked two or three months out, because everybody gets them regularly, so there's much higher demand for that service.
Re: (Score:2)
Availability varies based on whether or not you already have a doctor and other factors like location and income. If you don't have a GP currently, you better get one. I've had the same GP for ~15 years and his medical group stopped taking new patients ~5 years ago. There's no waiting list, it's just closed to new patients until the patient pool "shrinks organically" or they get more doctors and staff to take on more patients.
Re: (Score:2)
It's utterly hilarious to me that this is sold as a sort of a negative.
Meanwhile here in Finland attempts of getting "your own doctor who would know your medical history and therefore would be more effective in treating you" has been at least a 20 year project. Probably longer.
It failed so far. Every time. I have "own doctor" in the documentation. Out of my last ten visits, I think I got to see him once, if at all.
Which is incidentally my point. There are many things that Americans take for granted in their
Re: (Score:3)
No we're not, or at least those of us who bothered to spend a few years serving our country when we were young. Then, when we come home from whatever branch of the Military we served in, we can sign up with the VA and get inexpensive medical care for life. And, before you start hunting down horror stories about bad care or bad conditions in some VA hospital, take a moment to think about similar stories you've se
Re: (Score:2)
There is nothing stopping you from going to the US to visit a doctor. The US system allows it. The flight there will be less than the doctor bill, so its not even that hard if you are the kind of person who can afford it in the first place, which it sounds like you are.
Re: (Score:3)
That's what, 6% of total population? Granted upper classes no longer really serve, so lower middle is going to be somewhat overrepresented, so let's double that number. It's still just 12% of the cohort.
Re: (Score:2)
Re: (Score:2)
Many people do. Top tier health tourism is a thing in US.
Re: (Score:2)
Now apply this same logic to my earlier point.
Re: (Score:2)
So, stop whining and put your money where your mouth is.
Re: (Score:2)
I do. I don't demand super cheap health insurance that allows me US levels of access to healthcare.
Re: (Score:2)
My mistake. When you said you waited in queues or whatever I assumed you were saying that was undesirable. Now that I know you actually like waiting in queues, this whole thread makes a lot more sense.
Re: (Score:2)
Do they train your kind to be this obtuse, or are you naturally like this?
Re: (Score:2)
Have you checked your vitamin D levels lately? You aren't making much sense.
Re: (Score:2)
I've never waited 6 months for a dentist as a US citizen. In fact, I was annoyed at the 2.5 months I use to wait as I was on an HMO dental plan. Now, I've switched to a PPO dental plan and I can see a dentist in under a month, easily. They actually call me every 6 months to schedule an appointment that's typically a week or two after that phone call.
I live in San Diego, so maybe we have more dentist practitioners then the average.
Re: Why can't they just ask... (Score:2)
Re: (Score:2)
Yes, this is the standard "I've been to foreign quarter of one of the T1 cities. China is great and so accessible to foreigners!" story.
Reality: You visited carefully curated face of China aimed at foreigners. You never so much as glimpsed into anything that ever resembles China that locals see. That was specifically blocked from you. Everything about your experience was curated specifically to isolate you in that bubble, and make you feel at home.
Fun part: did you ever notice your handler?
Re: (Score:2)
Why is it so difficult to see a doctor?
Privatized Healthcare is why!
You live in a cave? The news is 24/7 on UHC CEOs and Super Mario Bros.
Re: (Score:2)
Wow, that's a head-turner for me to read. In the state where I live doctors are very much in short supply.
For all focus on expanding healthcare at the moment (with the murder of the healthcare CEO) this is something that should be obvious but which isn't being addressed - let's say we raise taxes and pronounce that everybody has healthcare now. Where does that healthcare come from? There isn't suddenly a larger supply. Getting t
Re: (Score:2)
Re: (Score:2)
I'm a United Healthcare customer. My doctor visit claims get auto-denied.
Re: (Score:2)
Re:Why can't they just ask... (Score:4, Informative)
The higher the quality of the training sets, the better the AI will perform.
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
Its not a drain on infrastructure and resources. To get the same results from a human, you need to feed, clothe, house, educate, and supply that person. All of that takes orders of magnitude more than the current batch of AI, even as bad as it is. Its probably less resources overall to ask an AI to make concept art, and pay the human concept artist UBI to stay home and consume as little as possible.
Re: (Score:2)
I just asked ChatGPT how LLMs are useful and then how much of that answer was marketing bullshit. Here are the results:
1. Natural Language Understanding and Generation: With limitations in terms of context, precision, and the handling of ambiguity.
2. Information Retrieval and Q&A: Mostly accurate for general queries but prone to inaccuracies.
3. Automation and Efficiency in Workflows: Results may need fine-tuning and human oversight, particularly for technical tasks.
4. Personalization: LLMs are good for
They are not super intelligent yet (Score:2)
So you're basically asking advice to a 6th grader
Re: (Score:2)
What are essentially elaborate Turin machines are merely simulating human intelligence. They're not actually thinking.
Stating the obvious (Score:5, Insightful)
Re: (Score:2)
Its Turtles all the way down!
We fired everyone and replaced them with AI. We got the answer! Now we need to train experts. Then hire them to find the problems for the answer!
Re: (Score:2)
Unlike the blockchain hype, there is actually substance and real-world usefulness behind the "AI hype."
Kazakhstan, Bhutan... (Score:1)
You should be training the model on world-class specialists, not 3rd world specialists. Cost is not everything. Quality matters.
Re:Kazakhstan, Bhutan... (Score:4, Informative)
I realize that nobody seems to know what "third-world" means, anymore, and I expect that you probably also say "rizz" and "drip". The third-world countries were the ones that stayed out of the Cold War, which seems eminently sensible.
Re: (Score:2, Informative)
I realize that nobody seems to know what "third-world" means, anymore, and I expect that you probably also say "rizz" and "drip". The third-world countries were the ones that stayed out of the Cold War, which seems eminently sensible.
Originally, this terminology seems to have come from the United Nations. Which 'world' a country belonged to was defined according to gross national product, social and political factors. The 'First World' was the wealthy, industrialized, democratic nations, aligned with the USA world wide. The 'Second World' was the modern, wealthy (enough to compete militarily with the USA), industrialized, communist countries aligned with the Soviet Union world wide. Most of the rest of the world was deemed part of the '
Re: Kazakhstan, Bhutan... (Score:2)
'Third World', is nevertheless still used as an insult
Not so much any more. We've moved on. Now we call them shitholes.
Re: (Score:2)
Re: (Score:2)
I've always thought of 1st/2nd/3rd world as national GDP or maybe citizen income levels. Most of the "West" would be 1st world since we have high income, all the latest toys and bells. 2nd world is okay but less income, less toys but still decent enough.
3rd world I think of as a country whos average population doesn't have reliable electricity, water or food or Internet. Those things just "are" in a first world country and even in a 2nd world country.
Obviously, this definition doesn't really hold up when yo
Your ignorance is only matched by your smugness (Score:2)
I realize that nobody seems to know what "third-world" means, anymore, and I expect that you probably also say "rizz" and "drip". The third-world countries were the ones that stayed out of the Cold War, which seems eminently sensible.
Most who were "involved" in the cold war were merely trading partners with the developed world. Nothing about being behind in development is sensible. You're spending a lot more for basics, lacking in industry, paved roads, modern medicine, inexpensive food, etc. You may think life is better without McDonalds, but what about without advanced treatments for cancer or even basic things like heart/kidney disease? What about having shitty roads that cost 10x as much to build? Nothing is good or sensible a
So it was humans all along (Score:3, Insightful)
Re: (Score:2)
To all those who fear that AI will "take over" and we will all be ruled by AI, this principle is profound. AI is a creation of...people. It does what people instruct it to do. So yes, AI is people.
This is truly amazing (Score:1)
Re: (Score:3)
what about the 60-80+/hour weeks for only 80K part?
Being a doctor (Score:2)
Is the bulk of the job just remembering a whole lot of stuff?
Re: (Score:2)
Depends on specialization, but for GPs it's that plus collating stuff together. For radiologists and such it's mostly pattern recognition.
Re: (Score:2)
Harambe died for our sins.
They Need To Specify Type Of AI (Score:2)
Generative AI vs Predictive AI
Generative AI is creepy pseudo creations.
Predictive AI is pattern matching and data extrapolation. -- this is the useful industrial stuff
Re: (Score:2)
Generative AI vs Predictive AI
Generative AI is creepy pseudo creations.
Predictive AI is pattern matching and data extrapolation. -- this is the useful industrial stuff
Indeed. And the useful industrial stuff rather obviously benefits from having some actual subject-matter-experts involved in the projects done there. Note that LLMs are _not_ predictive AI, but generative AI and hence are basically unfit for professional use.
Hiring a lot more DOCTORS, maybe? (Score:2)
Editors, do your duty, or at least your due diligence...
Re: Hiring a lot more DOCTORS, maybe? (Score:2)
Or perhaps they meant "doctorates" (people with post-graduate degrees). TFA covered topics well outside medical domains (geospatial maps, for example).
This is the way (Score:2)
AGI is literally a poster boy. It's a pipe dream that isn't necessarily all that useful. The past, present, and future of AI is in domain-specific use cases. This is where the money has been made in the past and will be made in the future.
The discovery of new viable use cases will be the single most important area of AI progress. Of course, this will also necessarily require progress in AI models, including new architectures and paradigms.
In a way, the push for AGI is like space exploration. There is limite
Re: (Score:2)
I disagree. AGI would be extremely useful, even if slow and not very smart. However, there is no known way to generate AGI and not even a theory that could work in practice. No known AI approach can do AGI and that is a fundamental limit.
Or not (Score:2)
That "secret" only means that to be successful in an application domain, you need to understand that application domain. Hardly a "secret". More like the all-too-rare and elusive "common sense". Obviously, these experts cannot fix the fundamental defects of LLMs though. They can help with _other_ AI technologies.
In other words... no use case (Score:2)
They need to hire more folks with Ph.Ds, to see if *anyone* can come up with a use case for all the money they wasted, er, spent, on typeahead write large.
Re: (Score:2)
One of those comments that's going to age like milk. Maybe in the future AI will help write better posts.
Re: (Score:2)
Show us the existing use case, that actually works. Replacing call centers doesn't count.
Return of Prolog? (Score:3)
"Companies like iMerit and Centaur Labs are recruiting specialists worldwide, from radiologists in Kazakhstan to agricultural experts in Bhutan, paying premium rates for domain expertise..."
Sounds like an expert system from the 1970s and 1980s. So what's next, a resurgence of Prolog?
They've discovered SMEs! (Score:3)
We do the same thing in every area of software, not just AI. If you create mortgage processing software, you need mortgage experts to tell the developers what you want the software to do. AI is no different.
Experts from Kazakhstan and Bhutan ... (Score:2)
Save money on training your AI. Obviously the fastest way to "profitability"... the only measure is success