On the 5th October, the initiative “France is AI” took place in Paris, scheduled for two days, this conference was an opportunity to discover the ecosystem of the artificial intelligence sector and to assess the challenges the future holds.
Basically, an artificial intelligence system uses automatic learning methods to fulfill vision, language, text recognition tasks and so on.
On recent advances in machine learning, some breakthroughs have been achieved on the hardware and the computational power with Mipsology’s FPGA boards, NVDIA’s GPUs or the Google TPU devices.
— François Gouilliard (@F_Gouilliard) October 6, 2017
Even though, some questions remain without answers:
- How many bits do we really need to store and compute to obtain accuracy on a request?
- How long can we keep on storing data and how can we reduce the energy consumption during the computation?
- Will it be possible to run an AI on embedded device like a smartphone? If not, how to split the complexity of the computation?
In this perspective, Microsoft places its bets on quantum computing where others scientists like Francis Bach (INRIA/ENS) focuses on the importance of analyzing the capabilities of algorithms by more theoretical approaches ; or even Damien Querlioz who is developing intelligent memory chips that will “natively perform both memory and computing functions”.
— Leyton UK (@LeytonUK) October 5, 2017
The most important question is whether one day, we could trust artificial intelligence for medical diagnosis, interpretation and life or death decisions. That is the reason why it is important for the future of artificial intelligence to define and work towards the establishment of ethical rules on robotics and artificial intelligence.