Eric Schmidt has delivered a powerfully optimistic view of the world and the accelerating impact of technology at the Viva Technology conference in Paris.
Unless you’ve been living under a rock, you’ll know our industry is in a bit of a dilemma. As we continue to build smarter, faster AI, there seems to be growing concerns that computers will automate such much of what humans are currently employed to do, that there’ll be no jobs for large masses of the population.
as the Executive Chairman of Alphabet Inc (Google’s parent company) Schmidt spoke about Google’s recent investments in AI and talked in a surprisingly refreshing level of technical detail for a broad-ranging conference like this. Schmidt says the hardest computer math problems will have far reaching benefits to society, but the AI problems are so complex we need at least a 10x and maybe even a 100x performance boost from the best available today.
He spoke to the recent Tensor processing unit announcement which uses matrix multiplication, where the chips and the problems are both optimised to deliver the result as fast as possible by today’s computing limitations. He then spoke to the recent introduction of Auto ML where machines don’t simple use one neural networks to learn, but instead run multiple computer simulations that then inform each other the best method, ultimately computers, teaching themselves how to get to the finishing line faster.
At around this point, most people start to get fears of robots and machines using AI to make us all redundant, but Schmidt disagrees and is optimistic about the future with machines.
Computers are unlike humans in that they can efficiently learn from each other. Think about the information transfer that needs to take place for a specialist in their field to fully train another person, it takes years. Computers can do that transfer in seconds.
Schmidt used the example of training 2 cars. If you train the first to only turn left, then train the second to only turn right. They can transfer information between them and then both be able to turn left and right. For years we’ve understood that computers are better at performing math calculations and performing repetitive tasks, but many people never think about the opportunity when information transfer and learning can happen virtually instantaneously. Ultimately this means there’s a foot pressed firmly on the accelerator and technology innovation that affects all industries is about to accelerate, not slow down.
Schmidt says that in proposed automation tasks, research is showing that in reality of the tasks that can be automated, only about 90% of the task can be done by computers. There’s still a need for humans to play a role and rather than replace their job, Schmidt says jobs are likely to change to be augmented by computers, leveraging their abilities where it best makes sense.
When you think about a business that embraces this, their employees + computers could dramatically outpace the productivity available by simple humans alone.
I have to say, I tend to agree with Schmidt. Humans are amazingly creative creatures and thinking of robots, AI and technology in general as a productivity super power, may actually leave us doing more of the type of work we enjoy and let the robots take care of the boring, repetitive, or mathematically intense tasks.