Blog overview

Part 1: AI and Me

Diary of an experiment

KI-UW0

Uwe Weinreich, the author of this blog, usually coaches teams and managers on topics related to strategy, innovation and digital transfor­mation. Now he is seeking a direct confron­tation with Artificial Intelligence.

The outcome is uncertain.

Stay informed via Twitter or linkedIn.

Already published:

1. AI and Me – Diary of an experiment

2. Maths, Technology, Embarrassement

3. Learning in the deep blue sea - Azure

4. Experimenting to the Bitter End

5. The difficult path towards a webservice

6. Text analysis demystified

7. Image Recognition and Surveillance

8. Bad Jokes and AI Psychos

9. Seven Management Initiatives

10. Interview with Dr. Zeplin (Otto Group)

 

Should I risk it? I think I have to. Otherwise, I will come across as many others do: All talk and no action. I've frequently told conference and seminar participants that digitisation and, in particular, artificial intelligence will fundamentally change our world, and especially its economy, but do I understand what's really behind it? I'm aware of the theoretical concepts, possible failures of use, options for business models, the Internet of Things (IoT) etc. and can talk about them with some ease. I've also been involved in projects where AI was used. But I've never really peeped behind the curtains. That must change immediately.

I want to know what's possible, and how it works. How complex is it to develop with AI? Which components can companies use to create new solutions?

The journey began with a friendly invitation from Microsoft to take part in a free MOOC (Massive Open Online Course) on artificial intelligence and machine learning, with useful practical tips on programming units that one must master.

My Investment: Time, and a considerable part of my brain's processing power

It should take three or four hours a week. That could be tricky. On top of that, I'll have to program something myself. OK, during my studies I had to learn C, and in the mid-90s I got to grips with internet scripting languages, but I'm no seasoned developer. This means I'll probably have a bit of stress to deal with while learning Python, a language hitherto unknown to me. Admittedly, it does say on the course's website that a high school graduation and a certain amount of mathematical knowledge will be sufficient, but these terms are elastic. To be honest, right after high school I was much better at such things than I am now. So this is a good chance to either learn something new in middle age or fail spectacularly. He who preaches to others that one must have the courage not to stand still must also take the plunge himself. Let's go. I'll keep you informed.

The crux of Artifical Intelligence

What is intelligence? And what is meant by artificial intelligence? Experts have been arguing about the definition of natural intelligence for over a century. Its defining characteristics are often said to include abilities in logical and rational thinking, as well as problem-solving. In 1912 William Stern invented the concept of the intelligence quotient (IQ), and therefore introduced the world to a concept that made it possible to classify people, measure them against one another, and, for some, to cultivate a certain arrogance.

Sadly, the concept of IQ is often misunderstood. Of course it's more desirable to have a high IQ than a low one. But how do those with an IQ of 70 differ from those with one of 130? The only thing that can be said with any degree of certainty is that there are exactly as many of the former in the population as the latter, because the IQ level says nothing more than how many people there are with a particular level of intelligence. What is being measured is the distance from the established median of 100. An IQ of 30 points above this value enables entry into Mensa, since only 2% of people fall into this category. At 30 points below the median, according to the International Classification of Diseases, you've just about managed to avoid the so-called "mild intellectual impairment".

This doesn't mean that those with an IQ of 140 are twice as intelligent as those with one of 70. The differences can be slight, or so drastic that the two aren't worth comparing, or anywhere in between. Neither is IQ a reliable predictor of success: there are highly intelligent unsuccessful people, just as there are successful people with less intelligence.

Now the computers and the robots want to participate

Until now it has been very easy to classify the performance of computers: generation of processor, speed, memory, bus system, benchmark tests - all of this is empirically measurable. But when one enters the world of intelligence measuring, pronouncements on the subject become just as vague and hard to measure objectively as with people. During a recent lecture at Potsdam's Hasso Plattner Institute, Volkhard Bregulla (VP Manufacturing Industry & IoT at Hewlett Packard Enterprise) classified the current level of AI's IQ as being between 50 and 70: not irretrievably stupid, but a world away from what we consider intelligent.

IQ-AI

From a Human Resources perspective, AI therefore would qualify for extra classes after school, although the unemployment office could probably be avoided. AI wouldn't pass an assessment programme for higher management functions, even if some systems are already able to defeat, for example, chess champions at their own game: AI does not yet display sufficiently broad intelligence. We can, however, assume that AI will become more intelligent quickly, much more quickly than biological organisms can achieve through evolution. According to Bregulla, 2029 will be the year that an AI is created which is cleverer than every human. Eleven years: that's not a long time at all. We really should prepare.

No-one knows what it will be like - but we can shape it

The scenarios of how our world will look are legion. There are the cheerleaders at Singularity University , just as there are the critics, such as Warner and the authors of the Malicious AI Report. We will probably see very soon the consequences of AI, and the effect it will have on our everyday lives. While natural disasters, epidemics and mutations in organisms affect us more or less at random without our input, with AI we have one decisive advantage: we can shape it. That opens up the possibility that it can be a force for good, and indeed its potential is already being seen in the fields of production and digital services.

Microsoft's Azure platform is just one developmental environment for machine learning and AI among many. Similar services are offered by Amazon Web Services (AWS), Google Cloud, IBM Watson and of course specialised IoT platforms with AI functionality. It's just chance that I landed at Azure.

So it's off to the machine rooms of AI and the Internet of Things. I'll see what I can learn there. And I'll keep you informed.

 

next blog entry ⮕

published: June 11, 2018, © Uwe Weinreich

new comment


privacy