AI Section in forums

[radderfire]

All of which is achievable without "AI"

In a few weeks or months longer.

 

[Grim5]

To those with the skills, but those without it's a god send.

That's a bit of a problem then isn't it, you need to be skilled and know what you're doing to use the AI in the first place. The scary thing is dumb people asking questions, AI giving them rubbish wrong answers and them thinking the answer is correct because they don't know any better

 

[Feek]

I find it astonishing to hear how many people being dismissive of AI, and relegating it to just a fad only useful for making some terrible YouTube shorts.
I'm a GP and with very little effort it has helped me: get a domain, set up PfSense plus tools like HAProxy and Cloudflare to self host a site that uses LangChain plus a very modest use of the OpenAI API service to deliver up-to-date medical journal papers and local / national guidelines to a groups of several hundred clinicians in my area. My colleagues can simply upload the latest knowledge, the system ingests and then stores the info in a vectorised database - which uses a mobile friendly UI to answer clinical questions submitted by others. We can now access thousands of papers and guidelines in seconds. All because of AI, and without AI this would have not been possible. I guess what you get out of it is a reflection of what you put into it.

Is this AI answering medical questions?

 

[radderfire]

That's a bit of a problem then isn't it, you need to be skilled and know what you're doing to use the AI in the first place. The scary thing is dumb people asking questions, AI giving them rubbish wrong answers and them thinking the answer is correct because they don't know any better

Although the same can be said of traditional search. It's only as good as the resources it can scan.

 

[Rroff]

That's a bit of a problem then isn't it, you need to be skilled and know what you're doing to use the AI in the first place. The scary thing is dumb people asking questions, AI giving them rubbish wrong answers and them thinking the answer is correct because they don't know any better

Having this with things like YT comments, etc. and people based on their use of AI very confidently rubbishing what someone has posted, but are actually very wrong because they don't actually understand what AI is telling them, how to cross-reference to check it, or how to leverage AI to get the correct information.

 

[bazaarboy]

All of which is achievable without "AI"

exactly the sort of dim-witted response I fully expected hahaha -- "adding numbers is achievable without a calculator", "I can get all my news in a printed newspaper" etc Why would I choose to spend months slowly learning a multitude of programming and IT disciplines which I would seldom use, when AI helped me achieve ALL of it over one weekend?

 

[bazaarboy]

Is this AI answering medical questions?

no, medical literature is dense and there's lots of it - by ingesting & vectorising you essentially get to the answer quicker. It's not there to decide for you, but to present the information in a quicker and more efficient way. Say I want to find out the referral criteria for suspected myeloma, finding the full guidelines, scanning though the >100 page document and reaching the answer would take 2-3 minutes. With the tool I created its 5 seconds or less.

 

[D.P.]

exactly the sort of dim-witted response I fully expected hahaha -- "adding numbers is achievable without a calculator", "I can get all my news in a printed newspaper" etc Why would I choose to spend months slowly learning a multitude of programming and IT disciplines which I would seldom use, when AI helped me achieve ALL of it over one weekend?

And as a user of the system, sure anyone can spend weeks reading dozens of papers, doing hundreds of keyword searches, skimming hundreds of abstracts and taking notes on everything. Or they can ask a question and get the answer straight away with links to all original articles.

You don't need a vehicle to travel from Edinbough to London, but hiking the whole way is just not an efficient use of time.

 

[Armageus]

exactly the sort of dim-witted response I fully expected hahaha

much like AI, rubbish in -> rubbish out.

You literally said

without AI this would have not been possible

Clearly it is possible without "AI". Even the actual data analysis could have been done without "AI" - things like Elasticsearch and other Full Text based search algorithms have been a thing for ages.

Why would I choose to spend months slowly learning a multitude of programming and IT disciplines which I would seldom use, when AI helped me achieve ALL of it over one weekend?

Why would you have to? Guides and blogs on things have existed long before AI, and allowed people to achieve things without having to slowly learn anything.

 

[adidan]

no, medical literature is dense and there's lots of it - by ingesting & vectorising you essentially get to the answer quicker. It's not there to decide for you, but to present the information in a quicker and more efficient way. Say I want to find out the referral criteria for suspected myeloma, finding the full guidelines, scanning though the >100 page document and reaching the answer would take 2-3 minutes. With the tool I created its 5 seconds or less.

And you accept any answer as 100% correct?

Interesting.

 

[bazaarboy]

And you accept any answer as 100% correct?

Interesting.

the output has been programmed to come, as an excerpt, from the inputted information only, in other words no part of it is "inferred". Furthermore the excerpt is turned into a href link, so that you can immediately open the full text at the correct location.

 

[radderfire]

There's nothing especially intelligent about AI, it's just allowing people to use natural language to make queries, and to provide answers by using linguistic analysis. This is clearly a time saver and provides "answers" more quickly. It is also true that these "answers" cannot be relied upon to be correct, because you often won't have a trustworthy source of information, or you are analysing incorrect information.

It's a tool to get to an answer more quickly, but ultimately a human should be coming up with and verifying the correct answer. The danger is when no verification is done.

 

[Penfold101]

There's nothing especially intelligent about AI, it's just allowing people to use natural language to make queries, and to provide answers by using linguistic analysis. This is clearly a time saver and provides "answers" more quickly. It is also true that these "answers" cannot be relied upon to be correct, because you often won't have a trustworthy source of information, or you are analysing incorrect information.

It's a tool to get to an answer more quickly, but ultimately a human should be coming up with and verifying the correct answer. The danger is when no verification is done.

True AI would be like Ultron - wouldn’t be a fan of asking him medical questions…

 

[NickK]

Thise that have benefited from AI are those genuinely applying the tech rather than being a glorified search UI, or those that had such neolithic corporate processes that just needed investment to automate. The former will get ROI, the latter will fail because they are still continuing to drive the now-automated Neolithic business model.

 

[ANDREW GREEN]

I always was so annoyed at my parents for not investing every penny they had into amazon. This new company that everyone was talking about. All of sudden they were no longer going into towns but buying everything form amazon. How on earth did they not invest their entire savings into a company like that.

Now I look at ai. Isn't it the same thing? To stop using Google but use ai instead is a massive change of habits.

Remember Last Minute.com? Remember the Dot Com bubble? AI isn't making profit yet. There will be mega companies and mega flops.

 

[D.P.]

Remember Last Minute.com? Remember the Dot Com bubble? AI isn't making profit yet. There will be mega companies and mega flops.

The main difference is that in the dot com era most companies were not even generating revenue, while the current AI "bubble" is led by highly profitable companies that are investing sustainably into AI and where most of them are generating revenue, but at a loss. Costs per token have dropped massively.

For sure many smaller companies and startups will fail as they have jumped on the bandwagon without a sound business model, but the big payers are profitable overall, but AI may not be today.

 

[unwashed potato!]

Remember Last Minute.com? Remember the Dot Com bubble? AI isn't making profit yet. There will be mega companies and mega flops.

Sure but as the other user has said it's a different situation. The dot Com bubble was simply no one having any clue what was actually valuable and so worthless companies being sold as valuable.

AI is a different situation, and almost entirely funded by huge tech companies. I have zero doubt that alphabet is going to go bust any time soon.

You have these massive billion to trillion dollar companies investing everything into ai, and getting the rewards already. While we think whether ai is a bubble or not

 

[Hagar]

And theyre doing it by ripping everybody off. Creatives, scientists, everybody. Patents no longer enforceable are now theirs.

 

[silversurfer]

 

[sityeuropa]

There's nothing especially intelligent about AI, it's just allowing people to use natural language to make queries, and to provide answers by using linguistic analysis. This is clearly a time saver and provides "answers" more quickly. It is also true that these "answers" cannot be relied upon to be correct, because you often won't have a trustworthy source of information, or you are analysing incorrect information.

It's a tool to get to an answer more quickly, but ultimately a human should be coming up with and verifying the correct answer. The danger is when no verification is done.

From how you are speaking, I know you are not into programming.
If you were and do coding, you would know it is not just a fancy search feature.

I can't describe how unbelievable AI is from a coding point. It is the greatest technological leap I've seen since I've been using computers.
Feeding disassembled C code from a compiled program in IDA; into AI and having it to create native ports in python for the functions is just insane and one example I have recently completed with it.
I've also wrote an Android app having the AI write all of it; this to control some custom hardware of mine. I just don't have the time to code any more or keep up with the framework changes. But AI has given me a workable app just by describing what I want.

So if you think it is just a fancy language search engine, please try your hand at coding with it and see the magic for yourself.


People I find that don't like AI:

Creative career people, artists etc for obvious reasons - Massive threat to jobs, understandable, but these types of people seem to be more entitled thinking throughout history they are the only ones with a career that has had the possibility of being replaced by technology.

Labour / Green voters, possibly because of the climate crisis, unions etc with job losses.

People who think using Google Search is indictment of how AI is as a whole, not realising you are probably using the most basic non reasoning AI model for instant answers.