Blue collar thumbs up. Interesting. So manufacturing real stuff due to decoupling works here. It will too for make in India
low-end, low value add to mid-white collar thumbs down but with caveats. First-level customer service if not third-level. A category of call centre jobs at risk here maybe. I want to see how this industry is going to respond.
How good of a teacher can AI be? That could be useful at the secondary level. Since it's mostly one way to students. But Q&A will need a human. In any case, a lot of corporate training used video like two decades ago. With real people talking instead of avatars. But someone had to design the course. That job the AI must do. How well?
After making a good first impression I still think people are hyping the potential. They love to say Game over before the game even begins
Like they are in the fortune-telling business
The article you shared clearly states AI fights better than humans.
In the lab. Where is the real-world implementation of it? Nowhere. How well will they survive battle testing? No data to work with yet. Just promise.
A truism in war is the man behind the machine makes all the difference. AI is touted to replace that man. Ain't happening. Not in our lifetimes.
Where is AI being used then?
AI is already in the battle field , Sensor fusion and DRFM are all based on Neural Network and Machine Learning and DSP processing, Rafale does it , Tejas does it , Su30MKI does it AI is not new to Military it is their since 90s , Only rebranded from Neural Network to AI. Brahmos itself uses lots of AI/ML for SeaSkimming , Nirbhay uses AI for Terrain Hopping , these are nothing new
Rebranded, heh Since the 90s. Do you see how long it takes to adopt? If you can replace costly sensors with cheaper ones using AI then that can work. But proving takes a long time. You are in effect making a cheaper calculator.
Everyone has been saying that fifth gen fighter planes will be the last manned fighter planes as humans are the bottleneck nowadays. Most missions these days are flown by drones (not fighter planes). RC or AI controlled drones are supposed to be the future.
Ya, this is what people were parroting after Azerbaijan-Armenia then Ukraine happened and we learnt drones are not game changers and are incapable of changing outcomes on the battlefield. They have niche uses. ISR and going after point targets. Would you allow a fleet of drones to be managed by AI for a mission? no.
Do you know what other myth got blown up since Ukraine? cyber.Same limitations. In a conflict at the border, it does not make one bit of difference if Bombay's grid goes down for a few hours.
The shocker that has come out of Ukraine is wars are not short affairs. They can be long and protracted. How sustainable are high-cost weapons systems in such a scenario? Himars system with its $100k missiles vs. $20 artillery shell. The West is scrambling to find stuff and can't induct too much high tech because training people to use this stuff takes many months. Simple is good. Complex is not. Battlefields impose constraints on tech that can be used. Glitches cost lives. Can't use too much high-tech up in the mountains.
It turns out people are cheaper than machines. Boots on the ground win wars. Sheer blood and guts. Tech assists where possible.
Why am I talking about war? because all the tech we take for granted today had its roots in the defence industry. This is where it's going to be used first and to advantage if possible.
Oh, at this level, github copilot can write better code than our indian grad students.
github is hobby level. I'm looking for more. How many engineers going to get fired at MS, Apple, Google or any of countless software companies? because you know AI can do it all now. I wonder what those employees have to say about such a prospect since they are all on the chopping block. All drawing six-figure salaries as well.
Which version of Android or iOS is going to be done entirely by AI ?
Which category of people is being laid off at companies like Amazon these days? support level or those developing Amazon's software. Who knows?
Creating something isn't difficult. Code generation for UI's was doable twenty years back. Then the real people took over to fill in the blanks.
I said maintainable. If you can't maintain it then it's useless. Sure these things can generate whizz-bang stuff but if a fix or modification needs to be done and the bot can't do it then a human is needed. And if they can't figure out what this thing has done you got nothing.
Code goes through a lot of iterations in its life cycle. It's constantly changing. Features to be added. Bugs to be fixed. Complexity to manage. These are all real-world problems which I see not much evidence for other than speculation.
So managers, employees working for me don't have mind of their own?
Why add to the problem? The idea with machines is they aren't supposed to have these issues. They don't get tired, they don't argue they are cost effective. Here is a problem get it done within these constraints.
There's an AI that edits your video, another AI that adds visual effects, AI that writes blogs, Or makes music. Just now there's an AI generated song by Drake and the weeknd going viral. Just get in the AI news.
Proofs of concept.
Na. What you are looking for is spoon feeding.
Just open the youtube and search for AI. Check the 2 minute paper channel or the one suggest by
@psyph3r
Don't I do that here already? How about some QPQ. Why should I have to look at claims people make here? I'm not going to do their homework. Not my job to prove negatives. If you can't answer the simple questions asked then why should I trust anything people say? Very little substantiation is given.
Am pretty sure only very few things being claimed here are going to see the light of day this decade. Expert systems were a big thing in the 90s. Who needs that costly expert when you can distil what he knows into some program and use that instead? Didn't quite work out and I'm supposed to believe something with a mind of its own will? LOL
in the CAD field, they allowed automated changing of parameters in complex mechanical designs. You change a part or its dimensions and the system works out what changes have to be made to the parts its interfaces with and so on and you get to see the impact of that change on the overall design. Better calculator.
Big things were claimed back then. A decade went by and I saw how slow the takeup was and what the impact was. Then the next wave came, rinse and repeat. This is going to be the same.
Just because you can replace people does not mean you should. I can hire cheaper workers to get a job done but if these guys do bad work that needs more checking, that has to be fixed or worse after it's not savings. Oh, and this lot has a mind of its own too. When was that ever an asset at the low level? That is what AI is replacing, isn't it? There are going to be loads of teething problems you won't hear about. Since the scope is much larger it just has to take more time to adopt than previous tech iterations.
What I'm looking for is, case studies where AI was introduced and the results explained and studied. Not lab stuff.
