The Future and Finance Forum on Artificial Intelligence was held in Zurich in the last few days, hosting a number of known researchers and experts in the AI field.
When asked about what comes after those specialized technologies that will enable us to do near-realtime analyses on datasets that took hours or days in the last years he replied with:
>In the next few years both algorithms and hardware will be become much better and open up even more analytical potential. This will accelerate more and more and then, with the developments in the last years [see Alpha Go that beat a Go champion years ahead of earlier projections] recent estimates from fellow researchers put the Singularity at 2029, a full 16 years earlier than what Ray Kurzweil predicted in 2005.
The more important question will be what happens at that time but honestly, nobody really knows.
While this might be a bit optimistic, it shows a tendency of narrowing in on earlier predictions as more knowledge about the way towards them becomes available.
What do you think about the current state of AI and developments towards the Singularity?
Just attended an AI conference in Zurich
You're a retard
AI and singularity are crucial to the dawn of Waifu Age.
Can't wait to become an Eternal NEET with the perfect harem of AI waifus to please me.
Open source / open hardware robot waifus when?
We fuck off from technology then and get real waifus, if there are any left, because by that time open source anything will likely not exist.
>wants to share his waifu with stallman
Stallman has jus primae noctis on your waifu.
Learning how to opencl a good start on how to make my own AI?
In 2016 deepmind did alphago, wavenet, dnc, lipnet and started to approach starcraft II.
What the fuck are people not at Deepmind even doing?
Different instance, no problem.
>What the fuck are people not at Deepmind even doing?
Manually improving the botnet, until Deepmind get the botnet to improve itself.
Then it is ogre.
no, learning AI is a good start.
I'm not going to write an AI with python
Why not?
Learning, and "real-deal" implementations are different,
so might as well learn with something you are familiar with, which is easy and fun.
>>Alpha Go
big deal. they got together enough GPU's to brute force go with a CNN
it was pretty cool.
fuck no
Enjoy your corporate propaganda... erm marketing show?
Google/Tesla Futurists:
>Singularity is Open libraries like TensorFlow and OpenAI will enable a new generation of ML capacity beyond what is current possible
>Advances in quantum computing and nano-engineering will enable use to solve even more complex issues
People who work on TensorFlow and OpenAI
>They're really only useful for classification problems
People who work on Quantum Computers
>IO to/from the processor is currently impossible
>Nobody has yet built a true Quantum Processor, just Quantum Annealing Processor
>Modern Quantum Annealing Processors are barely faster then simulations of the same work load, yet consume vastly more power and space.
People who design physical processors
>They aren't getting any faster bruv maybe more specialized.
>We're literally running out of room to make transistors smaller.
>Quantum Mechanics is a son of a bitch.
The next big thing is gene editing. But that's only 20 years out. Currently it is a GPU/FPU bound issue (with some consideration to memory bandwidth).
we're a long way from the singularity
hardware will need to improve in a massive way
i don't know anything about quantum computing, but all the claims look too good to be true. logic says you can't get something for nothing.
This desu
I don't want Google leading the development of artificial intelligence
singularity at 2050, minimum.
we probably won't even hear about it when it happens
lol, i've got bad news for you
your*
Where's that Miku image depicting object oriented programming when I need it?
I know they do, but I really don't want AI in their hands
well, hop to it then. research it yourself.
Jesus Christ, at last someone speaking the truth in this board.
I'll just add that DeepMind's papers are massive overstatements about mediocre results, and that the impressive shit that they show you is the result of months of computation on a cluster of GPUs... No shit they are impressive
>gene editing
What makes gene editing the next big thing? I've heard CRISPR is still a relatively primitive tactic, but even with improvement what'll actually be done with it?
call me when AI can find good solutions to NP Problems.
I have already learned that doing parallel processing in python is pita
In fact you don't. You need very low temperatures and very special conditions to not fuck it up. But it's real and it's coming..
Larry page wants to give basic income/life extension/robots to common people, he has stated this openly.
You should be glad that he's leading the advancement. Would you prefer jewckerberg?