4 ms·

There are lots of groundbreaking applications. But I’ll give you one that by itself justifies the technology. Early research shows that AI is able to detect ca

There are lots of groundbreaking applications. But I’ll give you one that by itself justifies the technology. Early research shows that AI is able to detect cancer 4 years earlier than a doctor traditionally would.

jonathankoren 3y ago
First, I love how “AI” is just a meaningless term here. The topic was generative text and images, and now you’ve brought up image classification, a different problem. But hey, computers! So let’s bring it back to generative systems. IBM Watson was a complete failure, generating nonsensical diagnoses and treatments. https://slate.com/technology/2022/01/ibm-watson-health-failure-artificial-intelligence.html https://slate.com/technology/2022/01/ibm-watson-health-failu...
- happytiger 3y ago
  > I'm gonna say something that probably some people won't like. I'm impressed by how much money the industry is moving by this so called "AI" movement and literally having zero concrete improvement on one's every day life (and by this I mean common people, not developers/techies). > I haven't seen, nor witnessed anything that you can benefit from all this AI hype. All that has done is opening the door for polluting the web with crap. And we keep on and on with the madness of training, wasting money (and resources) on gpus, generating tb of data. For what? To generate text, images? > If someone can actually point out a real example on the real world how all this madness helps to a better world, much appreciated, because honestly I don't see it. The comment I was replying to did not say LLM or Generative AI. It didn’t specify LLM and simply asked for examples where AI is making good. Instead of saying, “is there a source?”, or taking 2 seconds to Google “4 year cancer detection with AI” people are lighting up the comments about failed AI from IBM from ten years ago. Here is the source: https://www.unilad.com/technology/ai-detected-breast-cancer-years-before-developed-737052-20230325 https://www.unilad.com/technology/ai-detected-breast-cancer-... Also re Watson. Top Google article for the search terms “Watson medical ai” leads to this article, which contains the best response to your criticism: > I don’t think that the failure of Watson means that artificial intelligence isn’t ready to make significant improvements and changes in health care. I think it means the way that they approached it is a cautionary tale that lays out how not to do it. https://slate.com/technology/2022/01/ibm-watson-health-failure-artificial-intelligence.html https://slate.com/technology/2022/01/ibm-watson-health-failu... Whether LLMs deserve all this hype is debatable, but whether there is merit in AI technologies being used today is not. There are some pretty good use cases in medical, pharma and biotech, agricultural, etc. Few of them are based on LLMs but that wasn’t what the commented was actually asking. They were asking if “the AI hype” had any real applications, and it does. LLMs are just the manifestation of the Gartner hype cycle anyways. And we haven’t gotten anywhere near the terminus for them either, with almost every vendor and open source project planning to at least double model size in the next year as hardware and GPU capabilities expand. I think we have yet to see what LLMs can really do and will be much more pleased with them in 2-3 years as they mature.
10xDev 3y ago
They have been talking about AI being able to do better than doctors at cancer detection for over a decade now: https://www.wired.co.uk/article/ibm-watson-medical-doctor https://www.wired.co.uk/article/ibm-watson-medical-doctor
- happytiger 3y ago
  But it’s not Watson that delivered. https://digialps.com/the-first-ai-medical-device-that-can-detect-all-major-skin-cancers-just-received-fda-approval/ https://digialps.com/the-first-ai-medical-device-that-can-de... That’s like pointing to GM’s 1990s electric car while explaining that electric cars haven’t delivered.
  - 10xDev 3y ago
    Yeah I totally was not aware of that. We are specifically talking about being better than doctors. I said that word for word.
lm28469 3y ago
> AI is able to detect cancer 4 years earlier than a doctor traditionally would. Which "AI" ? Which "cancer" ? What's the false positive/negative rate ? Regardless of all that, if it detected it 4 years earlier it means it happened at least 4 years ago and have nothing to do with the LLM hype gp's criticising.
gwervc 3y ago
> There are lots of groundbreaking applications. > proceeds to give absolutely no example nor source Yeah sure.
- happytiger 3y ago
  I expected people already knew about this or would use Google. Here’s a link: https://www.unilad.com/technology/ai-detected-breast-cancer-years-before-developed-737052-20230325 https://www.unilad.com/technology/ai-detected-breast-cancer-...

jonathankoren 3y ago

First, I love how “AI” is just a meaningless term here. The topic was generative text and images, and now you’ve brought up image classification, a different problem. But hey, computers! So let’s bring it back to generative systems. IBM Watson was a complete failure, generating nonsensical diagnoses and treatments. https://slate.com/technology/2022/01/ibm-watson-health-failure-artificial-intelligence.html https://slate.com/technology/2022/01/ibm-watson-health-failu...

happytiger 3y ago
> I'm gonna say something that probably some people won't like. I'm impressed by how much money the industry is moving by this so called "AI" movement and literally having zero concrete improvement on one's every day life (and by this I mean common people, not developers/techies). > I haven't seen, nor witnessed anything that you can benefit from all this AI hype. All that has done is opening the door for polluting the web with crap. And we keep on and on with the madness of training, wasting money (and resources) on gpus, generating tb of data. For what? To generate text, images? > If someone can actually point out a real example on the real world how all this madness helps to a better world, much appreciated, because honestly I don't see it. The comment I was replying to did not say LLM or Generative AI. It didn’t specify LLM and simply asked for examples where AI is making good. Instead of saying, “is there a source?”, or taking 2 seconds to Google “4 year cancer detection with AI” people are lighting up the comments about failed AI from IBM from ten years ago. Here is the source: https://www.unilad.com/technology/ai-detected-breast-cancer-years-before-developed-737052-20230325 https://www.unilad.com/technology/ai-detected-breast-cancer-... Also re Watson. Top Google article for the search terms “Watson medical ai” leads to this article, which contains the best response to your criticism: > I don’t think that the failure of Watson means that artificial intelligence isn’t ready to make significant improvements and changes in health care. I think it means the way that they approached it is a cautionary tale that lays out how not to do it. https://slate.com/technology/2022/01/ibm-watson-health-failure-artificial-intelligence.html https://slate.com/technology/2022/01/ibm-watson-health-failu... Whether LLMs deserve all this hype is debatable, but whether there is merit in AI technologies being used today is not. There are some pretty good use cases in medical, pharma and biotech, agricultural, etc. Few of them are based on LLMs but that wasn’t what the commented was actually asking. They were asking if “the AI hype” had any real applications, and it does. LLMs are just the manifestation of the Gartner hype cycle anyways. And we haven’t gotten anywhere near the terminus for them either, with almost every vendor and open source project planning to at least double model size in the next year as hardware and GPU capabilities expand. I think we have yet to see what LLMs can really do and will be much more pleased with them in 2-3 years as they mature.

10xDev 3y ago

They have been talking about AI being able to do better than doctors at cancer detection for over a decade now: https://www.wired.co.uk/article/ibm-watson-medical-doctor https://www.wired.co.uk/article/ibm-watson-medical-doctor

happytiger 3y ago
But it’s not Watson that delivered. https://digialps.com/the-first-ai-medical-device-that-can-detect-all-major-skin-cancers-just-received-fda-approval/ https://digialps.com/the-first-ai-medical-device-that-can-de... That’s like pointing to GM’s 1990s electric car while explaining that electric cars haven’t delivered.
- 10xDev 3y ago
  Yeah I totally was not aware of that. We are specifically talking about being better than doctors. I said that word for word.

lm28469 3y ago

> AI is able to detect cancer 4 years earlier than a doctor traditionally would. Which "AI" ? Which "cancer" ? What's the false positive/negative rate ? Regardless of all that, if it detected it 4 years earlier it means it happened at least 4 years ago and have nothing to do with the LLM hype gp's criticising.

gwervc 3y ago

> There are lots of groundbreaking applications. > proceeds to give absolutely no example nor source Yeah sure.

happytiger 3y ago
I expected people already knew about this or would use Google. Here’s a link: https://www.unilad.com/technology/ai-detected-breast-cancer-years-before-developed-737052-20230325 https://www.unilad.com/technology/ai-detected-breast-cancer-...