CmdrLaForge
macrumors 601
AI tools have their limits, that get's clearer the more you use them. Still useful but you need to understand what you are doing.
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
ChatGPT's Apple Health Integration Flaws Exposed in New Report
- Thread starter MacRumors
- Start date
- Sort by reaction score
Yet, health insurance companies can accurately use AI to automatically deny approvals or claims with the sole goal of maximizing profits.
Séamus Boyle
macrumors regular
It’s pointless and so is the sleep score BS. Wildly inaccurate and a complete waste of time.
Apple's sleep score is far from perfect, but it's not bad. It's also highly helpful within person to track your sleep over time.
Edit: I keep meaning to write up an analysis of the sleep score and offer suggestions to Apple about how to improve it. I'm not a "sleep researcher" but have been on various peer-reviewed articles looking at sleep in different populations.
Last edited:
Ursadorable
macrumors 65816
How many billions of dollars did it take for them to get ChatGPT to learn to count the number of letters in a word? 😆
Then what are they for? More importantly, if they are not intended to provide at least semi-actionable insights, which again requires these to be accurate, why do you need user data to begin with.
A lot, although that’s not the only thing the tool can do.
Some people buy a $3000 MacBook Pro and then mostly just use it for media consumption and light browsing or other tasks. Other people use it to get a lot of computationally intensive work done. The same thing goes for LLMs. Some of us use them to get our work done much more efficiently.
You'd have to be crazy to link a black box 3rd party LLM to your health data.
LLMs are the future but they are still basically in a 'public alpha' stage, due to the immense cost of developing them i.e. they need subs and soon, adverting.
Oh and that a race is on of course.
Would I use a Health+service powered by Gemini though? Yes, I would because I would trust that Apple has tuned it correctly and that my data is secure (although we have no way of knowing if that's true or not - we'd just have to trust Apple).
LLMs are the future but they are still basically in a 'public alpha' stage, due to the immense cost of developing them i.e. they need subs and soon, adverting.
Oh and that a race is on of course.
Would I use a Health+service powered by Gemini though? Yes, I would because I would trust that Apple has tuned it correctly and that my data is secure (although we have no way of knowing if that's true or not - we'd just have to trust Apple).
Frosties
macrumors 65816
Ursadorable
macrumors 65816
You misunderstand my cynicism.
I know it's not the only thing it can do, but counting letters in a word seemed to be a massive hurdle for them that it took this long for them to make it work, tells me the I in AI is a lie. It can regurgitate what it's trained on with a fair degree of accuracy, but the lack basic intelligence of being able to identify the numbers of letters in a word tells me it's not very smart.
I am absolutely flabbergasted that the titans of an industry that has not released a valuable product is still going at it.
Jseeker
macrumors newbie
Same here. I noticed that using the sleep tracking, especially after they tweaked it for 'greater accuracy', I was getting more anxious from using it. So I turned it off.
One important thing that makes Google competitive is data. LLM is nothing without data. No company in world have so much data about humanity. So, to compete with Google, you need to harvest data.
asdfjkl;
macrumors 6502
It can't. They just trained it on the strawberry question. I just asked ChatGPT the following:
how many of the letter 'r' are in 'cranberry'?
There are 2 letter **“r”**s in “cranberry.” 🍒
Just wait until OpenAI tacks on enough cost to make a profit. Then you’ll get to pay for useless summaries!
PBG4 Dude
macrumors 601
This is the problem I see with AI today. If your cranberry question becomes as widely spread as the strawberry question was 2 months ago, models will be programmed to answer this question correctly. It doesn’t seem to be able to answer correctly without human intervention though.
Mousse
macrumors 601
This is what happens when you put reality TV personalities and Faux News anchors in charge of government services.
Just like how O'Brien can got Winston to see 5 fingers instead of 4, they can manipulate AI to include as many "r's" in strawberry as they want. You won't get the truth using AI; you'll get their truth.
Until we can make an AI that learns through experience--like we do--and not by being feed reams of data, then AI might be useful. The quality of AI we have now is reliant on the data it's feed. As the old programming adage goes: Garbage in, garbage out.
TwoBytes
macrumors 68040
There's a lot of criticism, but I'd like to see what the actual data that was fed into ChatGPT.
Garbage in = garbage out in most cases.
If the data had poor sleep scores (we all know how reliable they are) or years of missing data because someone upgraded devices or the goalposts changed on what counts as good data and bad data, then this could well have thrown off ChatGPT. EG If that dump included data that the person only slept 2h a night for 5 years, all of that has an impact.
Anyone who works with data knows it has to be cleaned and refined before processing.
Would I still like to see a real doctor? Yes.
At the same time, we're allgoing to use AI to get some sort of baseline and self-assess, and I'm sure those gaps will be closed in time after the beta.
Garbage in = garbage out in most cases.
If the data had poor sleep scores (we all know how reliable they are) or years of missing data because someone upgraded devices or the goalposts changed on what counts as good data and bad data, then this could well have thrown off ChatGPT. EG If that dump included data that the person only slept 2h a night for 5 years, all of that has an impact.
Anyone who works with data knows it has to be cleaned and refined before processing.
Would I still like to see a real doctor? Yes.
At the same time, we're allgoing to use AI to get some sort of baseline and self-assess, and I'm sure those gaps will be closed in time after the beta.
I've used Grok several times to review test results and treatment recommendations from my oncologist and others. In virtually every case, Grok delivered a detailed, fully customized summary of my reports in plain English. I wouldn't make treatment decisions based on AI just yet, but Grok has been a great tool to help me understand the medical jargon I've had to review. Like any AI, though, it's "buyer beware" regarding its feedback.
A reporter for The Washington Post has put ChatGPT's new optional Apple Health integration feature to the test by feeding it ten years of their Apple Watch data. The results were not encouraging, to say the least.
Earlier this month, OpenAI announced the launch of ChatGPT Health, a dedicated section of ChatGPT where users can ask health-related questions completely separated from their main ChatGPT experience. For more personalized responses, users can connect various health data services such as Apple Health, Function, MyFitnessPal, Weight Watchers, AllTrails, Instacart, and Peloton.
ChatGPT Health can also integrate with your medical records, allowing it to analyze your lab results and other aspects of your medical history to inform its answers to your health-related questions.
With this in mind, reporter Geoffrey Fowler gave ChatGPT Health access to 29 million steps and 6 million heartbeat measurements from his Apple Health app, and asked the bot to grade his cardiac health. It gave him an F.
Feeling understandably alarmed, Fowler asked his actual doctor, who in no uncertain terms dismissed the AI's assessment entirely. His physician said Fowler was at such low risk for heart problems that his insurance likely wouldn't even cover additional testing to disprove the chatbot's findings.
Cardiologist Eric Topol of the Scripps Research Institute was likewise unimpressed with the large language model's assessment. He called ChatGPT's analysis "baseless" and said people should ignore its medical advice, as it's clearly not ready for prime time.
Perhaps the most troubling finding, though, was ChatGPT's inconsistency. When Fowler asked the same question several times, his score swung wildly between an F and a B. ChatGPT also kept forgetting basic information about him, including his gender and age, despite it having full access to his records.
Anthropic's Claude chatbot fared slightly better – though not by much. The LLM graded Fowler's cardiac health a C, but it also failed to properly account for limitations in the Apple Watch data.
Both companies say their health tools aren't meant to replace doctors or provide diagnoses. Topol rightly argued that if these bots can't accurately assess health data, then they shouldn't be offering grades at all.
Yet nothing appears to be stopping them. The U.S. Food and Drug Administration earlier this month said the agency's job is to "get out of the way as a regulator" to promote innovation. An agency commissioner drew a red line at AI making "medical or clinical claims" without FDA review, but ChatGPT and Claude argue they are just providing information.
"People that do this are going to get really spooked about their health," Topol said. "It could also go the other way and give people who are unhealthy a false sense that everything they're doing is great."
ChatGPT's Apple Health integration is currently limited to a group of beta users. Responding to the report, OpenAI said it was working to improve the consistency of the chatbot's responses. "Launching ChatGPT Health with waitlisted access allows us to learn and improve the experience before making it widely available,” OpenAI VP Ashley Alexander told the publication in a statement.
Article Link: ChatGPT's Apple Health Integration Flaws Exposed in New Report
ChromeAce
macrumors 6502a
Oh yes bring us more articles about how software is failing during beta testing.
Also, the guy who got an F against the opinion of his doctor and insurer? Watch for the headlines that he dies of a heart attack unexpectedly. I’d put the combined wisdom of a chatbot up against the average capabilities of a typical internist any day.
Also, the guy who got an F against the opinion of his doctor and insurer? Watch for the headlines that he dies of a heart attack unexpectedly. I’d put the combined wisdom of a chatbot up against the average capabilities of a typical internist any day.