Microsoft Research's Chris Bishop: We give students university degrees for the same reason we scold AI

Chris Bishop, the director at Microsoft Research AI4Science, released a new book called Deep Learning: Foundations and Concepts, and while presenting the book in the Machine Learning Street Talk podcast, he had quite a few things to say about the ever-evolving AI technology.

He agrees that the general reaction to the capabilities of the artificial intelligence is not necessarily a positive one, even oi AI is indeed capable of providing correct information 95% of the time.

It’s interesting that we do seem to view the capabilities of neural nets with almost a different ruler to that of humans.

Chris Bishop

He stated that the reticence towards AI is due the technology becoming more similar to the human brain, and the human way of thinking things through.

I wonder if parts of the reaction that we have to these models is a little bit of that sense of that threat to the specialness that we feel as humans and I may be completely wrong, this is pure speculation.

Chris Bishop

He goes on to say that AI is not really that different from a top university student, yet we scold the former, while praising and giving degrees to the latter.

It’s interesting that people use phrases like ‘stochastic parrot’ [to describe AI]; it’s just regurgitating stuff that it’s seen before. Imagine there is a very smart physics student, went to a top university, worked really hard for 4 years. They would read books, they would read papers, listen to lectures, have discussions with their professors and with other students. And then they sit their final exam and they pass it with 95%, and they come top of the year. We don’t say ‘Well, 95% of the time, they’re a stochastic parrot regurcitating information, and the other 5% of the time, they’re hallucinating. No, we say ‘Congratulation, you have a fist class honor degree, you’ve graduated with honors.’

Chris Bishop

In many ways, what he’s saying is not exactly far the truth: Microsoft has been placing all the bets on AI for 2 years now, announcing Copilot, its popular AI model, coming to virtually every product and service, from Microsoft 365 to Windows 10, and Windows 11.

But the reception was somehow lukewarm. Despite the focus and emphasis on its capabilities, and the way AI can revolutionize work and how we get information, the adoption rate is slow. According to our survey, only 11% of those who can access Copilot use it for work.

However, the Redmond-based tech giant has been investing a lot of resources into AI research, and if you’ve been following us, we reported on quite some interesting AI studies in 2023: from Gorilla AI, which was the closes AI to reach AGI (the computation process that is comparable to that of a human mind), to WizardCoder, or Project Rumi (this AI can detect emotions and facial expressions and respond accordingly), Microsoft is interested in taking AI to the next level.

The company has seen partnerships with OpenAI, and Mistral, that will result in gigantic AI projects, such as the Stargate initiative, which will take the world by storm.

But maybe, the company should look for approval in places and situations it might not get it at all. Despite the capabilities, AI is not going to be widely adopted by people, but it can be repurposed in places where it could make a difference: the healthcare industry, for instance, which is already seeing the implementation of AI technology in the TRAIN initiative.

There might be a while until Chris Bishop’s words no longer can be considered the truth. But until then, AI companies should look into ways to use the technology to improve existing industries, not create an entirely new one that doesn’t appeal to the general public.