Hacker News new | past | comments | ask | show | jobs | submit login

>>Why only ML/AI? Because they are in demand now?

NO only that, but because it can be the new IT. Java, Hadoop etc are only languages and frameworks, that had at most a life span of a decade. But ML/AI is a whole new world. If things are going to be progressing in the current pace, then ML/AI could have a big impact on the world as the personal computers did and can be the next big thing of this century.

All this automation and job loss that the original article saying would be nothing if AI comes into full effect. There could be serious job losses all over the world. And with the biggest of population in the whole world, India would be the worst affected.. It is best for any country's interest to progress in AI so that they could have a say in future policy related to AI, not to mention getting more jobs related to that field.

Few of the several fields that could be affected by a combination of AI/Robots/Automation that I could think of ->

> Manufacturing / Factory workers > Defence personnel > Agriculture > Auto drivers (Taxi, Truck, Rail, Ship whatsoever) > Teaching (Virtual teaching) > Construction (Ready-made materials combined with smart machines) > Medical professionals (X-Ray diagnostics etc) > Doctors (though far-stretched but could be possible)

Why now? No body have solid answers when or will this happen, but if you look closely it's been only couple of years that machines got the ability to beat humans in at least few tasks we figured would require advanced intelligence, including accurate visual/speech recognition. That coupled with smart robots, and automation could replace a large percentage of existing jobs. Self-driving cars are a proof of this. This is little scary, but not alarming if there are proper policies in place. Any country need to be in the forefront of this technology or else can get sidelined really quickly.




>>But ML/AI is a whole new world.

For almost all programmers who will eventually work with this thing, they will mostly use some frameworks to do it. The real math behind all this stuff is quite old. There are already courses available to do that. Indian statistical institute offers beginner to advanced courses in statistics which you can take, they have been available for decades now.

Anything that you will likely do with ML/AI will likely involve using a framework today.

>>All this automation and job loss that the original article saying would be nothing if AI comes into full effect.

We don't even know if that kind of AGI is possible, let alone achieving it anytime soon.

Heck we haven't even come close to defining what AGI is.


>>For almost all programmers who will eventually work with this thing, they will mostly use some frameworks to do it.

Frameworks only provide part of it. In the case of AI, they just provide a general way to represent neural networks. Each use case need a different type of network with hundreds of parameters fine tuned and trained. Of course you can download pre-built algorithms and model zoos, but they are at best couple of years old.

Also do you think that the best things will be available under a framework? There are open source frameworks (like Tensorflow), but you will be surprised at the amount of research under progress and what portion of it is available for public. For e.g. look at the amount of work that China is doing in the field (And available for public scrutiny) - https://www.youtube.com/watch?v=xhp47v5OBXQ.

India (or any other Country) could do proper research in the area or end up buying these stuff from other countries, provided they give state of art technology to others. Just imagine a drone or an army of drones equipped with such AI technology can survey 24/7 and instantly take down your enemy even across a country.




Consider applying for YC's Spring batch! Applications are open till Feb 11.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: