no AI here!

Posted on Jan 28, 2008 by in artificial intelligence | 0 comments

“Artificial intelligence is not a term generally used at IBM.” (Kathleen Keeshen, IBM spokesperson, 1982)

Since the early days of software, the highest enthusiasms and deepest disillusions have followed the term of artificial intelligence when used by corporations, especially when included in, or close to, their product line.

Of course, the continuous misuse of the term, the total divergence in its interpretations (especially from customers and investors) and the continuous changes in its definition made many companies suffer because of their proximity to it, which explains the famous quotes of Kathleen Keeshen: we are serious people, not dreamers, therefore, we don’t do AI.

“We tend not even to use the term AI. The tendency now is for AI companies to embed expert systems into conventional products to build up a functional unit. They strive at becoming known as successful solution providers, rather than AI technology enterprises. A typical example of this covert approach is a sales support program for a retail company that takes orders by telephone. In its standard form, the program would simply check the presence of a requested item in inventory, record the sale, prepare the invoice, and advise shipping to act on it. The new-wave AI touch consists in “embedding” an expert system into this program. To the sales clerk it looks exactly the same, but for one exception: for sold-out items, suggestions for alternative choices pop on screen.” ( Harry Reinstein, president of Aion Corp., ~1988 )

More time-related testimonials like this one go in the same direction: in order to avoid being connected to the one specific focus of the moment, we don’t use the term, even if 2 years before (or 2 years later), its understanding would fit perfectly with what we do.

Another clear motivation of avoiding this term is due to the well spread-out understanding that “artificial intelligence doesn’t work”. Which doesn’t mean anything, but which most people agree with, especially when followed by : “And if it did, we would know about it, right?”.

However, all multinational companies have activities in AI (Microsoft, IBM, …), but: this is research. Now it is interesting to look at how they name it : while IBM clearly position themselves at leaders of AI since the early days, Microsoft calls it : Machine Learning, Adaptation and Intelligence, which, from a corporation of this size, is interesting. It is not only avoiding the word artificial, but is separating machine learning from Intelligence, as if it was yet already a more “serious” domain which deserved to get rid of the AI connotation.

In reality, there is a much bigger reason why AI can be a dangerous term to use, especially when describing your product line. As one of the most difficult tasks of a software company is to set customer expectation close enough to what is really delivered to them, the promise of a company should lead customers to understand why their dreams about the product they want should be substituted by the product you can offer them. In this logic, speaking about AI is the best way to make the customer’s mind fly far away to a place where nothing is real or tangible and where software can do anything because : “it is intelligent”. I personally noticed several times how people can quickly project their expectation of the intelligence of the software as much higher than what they would expect from a real human being.

All this considered, I have to agree that the position of the large companies is pretty smart, as they create the following effect in people’s mind : “This company is doing amazing things in research with AI and I must make them our main supplier because, the day when these things will work, I want to be the fist one served!”. Of course, this is just a continuous motor which is is always future oriented, regardless of what they projected to accomplish in the past. Unlike the release of a product, everyone understands that there cannot be any real release date in research, therefore you can easily, like Microsoft, announce the arrival of Speech Recognition in the next five years many times- no one will hold you responsible for it:

“I talked about some recent advances that really have me excited. Here are some that are literally within the next few years. Speech recognition. A big, big breakthrough.” Bill Gates, 16 November 1997

“Speech recognition will be part of the interface. That has been a tough problem because people are very demanding of high quality. The keyboard is not that bad. So, even though we are going to have this extra speed, probably we are four or five years away from that being the typical interface.”Bill Gates, 26 March 1999

“Likewise, in areas like speech recognition we have seen companies come and go. It’s a very tough problem, but it will yield itself to breakthroughs that will make that just common sense for every computer over the next five years.” Bill Gates, 17 April 2002

“Another big change you’ll see is that we’ll have microphones on PCs and the speech recognition will be built-in as a standard feature. And that’s probably two to three years from now that that really becomes mainstream…” Bill Gates, 14 October 2005

(If you are interested, there are many more, click here)

I would like to add a last interesting quote which shows how AI can be used to threaten people quite efficiently, but without taking any risks from your side. This is a pretty refined approach which, I think, will be proven more and more efficient in the future, as we clearly passed the line where customers have any concrete idea of what they buy, but just that they have been told they need it, that it looks good and that it is listed in Gartner.

“The ability to learn faster than your competitors may be the only sustainable competitive advantage.” Arie P. De Geus

Which can be used like this: without our products, you will take the risk to be soon behind your competitors on absolutely everything, because they will adapt faster than you. In a way, this means that AI would be the key ingredient for adaptation and survival in a Darwinist economy.

To conclude, it is clear for me that AI will always be a slippery term which will continue to cause great trouble to many companies when not used diligently. On the other hand, AI is the ultimate incarnation of what customers are expecting from a software product, and therefore, it is more than probable that it will continue to be used, for the best or the worst.    Send article as PDF   

Submit a Comment

Your email address will not be published. Required fields are marked *