Act of creation

Neural networks are the artificial brains that will solve all your computing problems before finally sidelining you.

Neural networks are the artificial brains that will solve all your computing problems before finally sidelining you.

Steven Thaler is no run-of-the-mill innovator. Where others have devised plans to simplify complex operations and introduce new business methods, Thaler's research into artificial neural networks (ANNs) has yielded a synthetic brain that not only thinks, but innovates, dreams and has near-death experiences. B&T chatted online to the St Louis-based academic about the future of everything.

How and why will artificial neural networks affect the life of the average IT director?
In the next 10-15 years, ANNs will quickly become a sine qua non for all IT directors, allowing them to model complex systems where no explicit knowledge is available to do so (as is usually the case).
Within this time frame, IT directors will be at their happiest as they and their organisations solve one previously intractable problem after another.
But within the next 50 years or so, IT directors will be much less happy, since ANNs will be capable of doing it all: the design, implementation and maintenance of other intelligent systems. In the long haul, IT directors will find themselves serving merely as a human-machine liaison, periodically eavesdropping on what is a completely autonomous machine community to be sure that humanity is still safe and well. It will be a lonely job, perhaps compensated for by an evening bartending job!

In terms of dollars and cents, what effect will ANNs have on the business world?
It's difficult to say quantitatively what the savings would be, other than to state the obvious: that cost savings will be significant and that ANNs will have the biggest savings impact of any other AI [Artificial Intelligence] paradigm.
As a rule of thumb, in my own neural network consultancy, AI solutions reduce costs by a factor of between five and 10.
The savings stem from my not needing to understand the nuts and bolts of the culture I'm working with. I only need some initial intuition about what determines what - that is, causality.
In this way, I can rapidly perform amazing feats like design a new toothbrush, compose new songs or navigate legal statutes for my client's gain. The crux of the matter is that I can do all these things, even though I'm not an expert in these problem domains. Herein lies the primary source of savings.

What effect will that have on those employed in a corporate or IT context?
Domain experts will no longer be required. A slight expansion in the number of your neural network-savvy IT staff means other parts of your workforce can be 'excused' - perhaps in order to study ANNs.
In addition, whole new avenues will open up for doing business. For one thing, with ANNs you can quickly glean who is contributing what to the bottom line within an organisation. This optimises the overall robustness of a business unit and the issue of career advancement is made refreshingly objective.
In other words, in the near future an employer may be able to assess raises and promotions on the basis of true merit rather than politics, bias or hidden agendas.

And other influences?
On a grander scale - and this is my more visionary part speaking - ANNs will have a profound effect upon the whole notion of capitalism. They will usher in an era where fortunes aren't made via the often shady manipulation of money but on who contributes what to the overall welfare of society.
Finally, in the coming era of "super-capitalism", the rewards will go to the brilliant MD who wipes out Aids, or the humanist reaching out to improve the global quality of life, rather than the classic fat cat calculating how far they can dupe the public.
I can proudly say that ANNs are the fundamental economic paradigm shift just waiting to kick in - and perhaps the basis of a new world currency.

When will all this happen?
The technology and the potential cost savings are here now. The rate-determining step is social inertia, as the older schools of AI protect their jobs and the public starts to appreciate the immense power of ANNs.
I predict (or perhaps fear) that the neural network revolution will come within the next five to 10 years, largely in the developing world, where there are no "anchoring heuristics", where R&D and resulting business may be conducted with a simple PC-based Lan. Challenged by these developments, the superpowers will accelerate ANN development and applications in the following decade.
Who will lose out?
We always thought it would be the blue-collar jobs that automation would first replace. Instead, the technology begins to supplant the white-collar 'thinker,' building ambitious models of very complex and ostensibly unfathomable problems. Be afraid. Be very afraid!

Is there any good news?
For every field of human endeavour made obsolete, 10 new ones will emerge. All it needs is a familiarity with what ANNs can do and not necessarily how they do it. In the case of IT programmers, for instance, it would be foolish to begin this millennium without acquiring a working familiarity with ANNs.
Outside the IT world, neural networks are just waiting to be discovered. They just have to be implemented in very autonomous packages that may remain opaque to the user. After that, imagination is the only limiting factor. Restaurant owners may intelligently adapt their menus to their clientele, and employers may more adeptly screen job applicants.

How and why did you get involved in this field of research?
While studying for my PhD in physics in the mid-70s, I had a tremendous interest in what are called Ising models and spin glass systems - essentially models of ferroelectric and ferromagnetic crystal lattices. These physical models are primitive neural networks in which the bi-stable switches are electric or magnetic dipoles that may abruptly change state. The connections between such atomic switches are physical interactions between atoms.
Later, while employed at McDonnell Douglas in St Louis, I used artificial neural networks to rapidly develop models of physical, chemical and engineering systems when no explicit models were available.

Claim to Fame: Steven Thaler

28 December 1949, St Louis, Missouri
Current position: President and CEO of Imagination Engines Inc (
What he's done: Produced neural networks capable of human-level invention, discovery and artistic creativity
What he says: Future IT directors will be reduced to machine minders

Read more on IT jobs and recruitment

Start the conversation

Send me notifications when other members comment.

Please create a username to comment.