Peter Cochrane: Is AI Schrödinger's brain?

Artificial intelligence today is barely intelligent at all. So what will it take for true AI to be developed, asks Peter Cochrane?

AI now seems to enjoy the duality of being capable of bestowing super mental powers upon us all, while at the same time being a force that may decide to enslave or even exterminate us.

This runs parallel to Schrödinger's immigrant who is too idle to work while at the same time stealing all our jobs. Both are a manifestation of a ‘post-truth world' where people appear incapable of deep thinking and analysis; where simple irrational beliefs are preferred over (often) complex verifiable knowledge and truths.

Dystopian futures are not just dreams, they are possibilities we might just realise. However, technologies are not inherently evil,. We are the ones programming and training robots to hunt and kill; we are the ones installing cameras to monitor behaviours.

At one extreme, China is looking to wide-spread conformity of citizens through real time behavioural analysis using cameras and AI. Credits are to be awarded to ‘good citizens'; thereby guiding them towards the ‘accepted' ways of society. Bad citizens can be denied credit, flights, train tickets and bankrupted.

It does not have to be this way! But avoidance is easier if people are educated and capable of rational debate based on knowledge and understanding. They have to see both sides with an ability to balance good against bad.

Sadly, this does not appear to be the case across much of the West where ignorance is now celebrated, and fools are built-up into heroes. Not knowing anything about technology is now something people are proud to declare and even boast about, while they also deride expertise.

The real human-AI dilemma is now:

"Switch it all off, and/or cease developments, and people will die";

"Remain oblivious to the potential downside and dystopias may result".

But are the projected downsides actually likely? Almost anything is possible for sure, but the result need not be so negative.

All we have to do is act responsibly. Witness the progression of AI in medicine, genomics, proteomics (the large-scale study of proteins, since you ask), logistics and industry, and then consider the graphic below showing near 100 per cent of AI systems on the bottom rung, almost entirely dumb, algorithmic, and only at the edge of intelligence.

We have many more rungs of the ladder to climb before AI can become a generalised, intelligent and self aware platform. Clearly the benefit opportunity space is vast.

This table illustrates how I believe artificial intelligence will develop. These projections and categorisations are independently created, but broadly line up with others in the field.

As I started from a different standpoint and adopted a different framework, arriving at a similar conclusions is reassuring. The transition years provided are my 'best guess' at the shortest possible times to change - reality may be longer.

The basis of the extended path to true AI is as follows:

Tenets

AI always seeks the ‘truth'

Axioms

Things that think want to link

Awareness

To be aware is to be alive

Suppositions

Do all intelligent/living systems enjoy awareness? At this stage we just don't know!

Scientists and engineers are nothing but thorough in establishing workable frameworks, models, theories and experiments aimed at verifying discoveries, claims and truths.

And we should take succour in the sheer numbers of people engaged in this field, because the reality is, we cannot stop, and we cannot turn back without some unacceptable human price to pay!

Peter Cochrane OBE is an ex-CTO of BT who now works as a consultant focusing on solving problems and improving the world through the application of technology