AI : The Next Fifty Years

Exploration and exploitation of AI is a very long-time adventure - thus, prediciting what may happen in the next 50 years would be a great challenge. Still, some actual developments, near-term hopes and long-term aspirations of the entire field are captured within one single view - Presented in Dagstuhl workshop in 2018>>

 

To come to that point, we capture the long and windy road in 3I = Invent, Invest, Increase - shown below ..

Comments, counterpoints and criticims MOST welcome !

SlideDeck >>

Kemal A. Delic

Martin A. Walker 

 

Tags
big data neuromorphic computing quantum computing AI high-performance computing

Kummenti

Profile picture for user n00236w3
Mibgħut minn Stefan Keller f’din id-data:Sat, 15/06/2019 - 19:10

Hi Kermal, 

I'm not sure that your slides are looking all that much forward into the future. Rather I think your topics look at the next 5-10 years.

For me the big 50 year topics are:

1.) To what extent will we become fully dependent of AI and automated deicison making?

2.) To what extent will critical decisions affecting society be based o AI?

3.) There should be some positioning with regards to the Singularity ideology.. 

Regards

Stefan

In reply to by Stefan Keller

Profile picture for user kad
Mibgħut minn Kemal DELIC f’din id-data:Sun, 16/06/2019 - 11:20

Thanks Stefan, 

This is projection of technology development and not social and economic impact

Which is by far more complex and highly unpredictable. I agree that we need to explore

Better impact of the web on individuals and societies, which is currently labeled as

Webology - web + sociology. AI will play major role in it - you are right..

 

Thanks for commenting

 

Kemal

 

 

In reply to by Stefan Keller

User
Mibgħut minn Christian RUSS f’din id-data:Wed, 10/07/2019 - 08:33

Hi Stefan,

agree with your points on the time & future perspective, but fhe full impact of quantum computing is probably more than a decade away, especially if we talk about mass market...

Best

Chris

In reply to by Christian RUSS

Profile picture for user kad
Mibgħut minn Kemal DELIC f’din id-data:Wed, 10/07/2019 - 09:35

Very much so Chris .. 

as QC theory is in long-time development, seeing QUANTUM COMPUTERS

materialized will take several decades .. also, it is most likely that the

VERY SPECIAL KIND of PROBLEMS will be solved by them. It would surely

be NOT office or accounting software. One could envison QC as the co-processor

to the current CPU/GPU hybrid architectures cracking on the hard/grand challenge

problems.

 

Thanks for commenting..

All the best, Kemal 

 

 

User
Mibgħut minn Aaron Turner f’din id-data:Mon, 24/06/2019 - 21:47

Hi Kemal

Many thanks for uploading your slides.

However, I would caution against the use of neuromorphic chips for AGI (= advanced AI) purposes (although I realise I am likely to be alone in this opinion!)

My reasoning is that (1) an AGI built using neuromorphic chips may be conscious, and (2) as a deliberate design decision, consciousness in an AGI system should be avoided, for both safety and ethical considerations.

Please see my brief note on the subject here.

Regards, Aaron.