The Promise of Artificial Intelligence

By George Ugras

Managing Director, IBM Ventures

There was a recent discussion in New York City titled, Should We Trust The Promise Of Artificial Intelligence?  The two sides argued passionately about how the creation of something in our image leveraging the continuous march of Moore’s Law and better networks and faster data storage and improving algorithms would lead to two very different versions of the future.  

Meantime in our daily lives we are already experiencing many improvements to our quality of life as a result of these technologies. Lives are saved by better weather forecasting, catching terrorists before they act, improved medical diagnostics, predictive maintenance on critical machines like airplanes. We are clearly at the beginning of a tremendous revolution.

We are clearly at the beginning of a tremendous revolution.

There are three technology components that existing and new businesses will need to thrive in this new world: dynamic datasets, pragmatic AI, well designed APIs.  Data continues to become the main gravitational force.   Here proprietary first party datasets in the cloud become more valuable, as both training sets and real currency (already true in advertising world, likely to expand to other segments).  In the consumer world there are currently ‘owners’ of this data who are exploiting these huge pools for mainly improved advertising, highly personalized and geographically and demographically targeted.

On the enterprise side things get more interesting as there are many disparate datasets that need to be incorporated.   An insurance company, for example, that is interested in developing a new offering may want to use behavioral and demographic data about their current customers which might be internal, third party data such as traffic or weather patterns, and open source data (from government and others) on population level statistics and real estate prices.

What types of new enterprises will be built is an intriguing question.  Recently a bank was launched in the UK that has all its assets in the cloud on day one – hence using new chatbots for them will just involve exposing an API.  Long term viability of these chatbots might be up for debate but the agility to distribute intelligence to the edge is obviously appealing.  We need to think of applications that extend beyond automated customer service and digital personal assistants.

How about launching whole new businesses that do not exist today?  Could advances in AI enable one to launch a new bank solely in the form of a chatbot?  If one can interpret 22,000 pages of Dodd-Frank regulation using AI without the gravity of employing hundreds of employees and decipher a customer profile from behavioral and other types of data this indeed may become a possibility.

We need to think of applications that extend beyond automated customer service and digital personal assistants. How about launching whole new businesses that do not exist today?

At IBM we are helping hundreds of clients and partners navigate this new Cognitive Era by offering platforms such as Watson and The Weather Company that encompass leading edge data ingestion, curation and cognitive capabilities.  We have also been actively enabling new types of businesses being formed by start-ups by opening Watson up to the world. There are tens of thousands of developers working on the Watson platform today.  The next decade is clearly going to be won by enterprises that internalize and excel in running on data rich cognitive platforms and retool their processes around such tools.

Going back to the moral discussion on AI – the audience’s view of AI being in essence dangerous and not trustworthy went from 30% to 59% during the debate, more a reflection of the pundits ability to manipulate the audience versus a result of a deeper understanding of AI.  It is hard to conceive a world in which we’ve built machines smarter and more creative than ourselves (success at a Turing test or Loebner prize is likely irrelevant to this future).

We will continue to evolve to function at higher level of abstraction in how we relate to machines as we’ve always done: we carried stuff, now we drive machines that carry stuff and soon we will be driven with our stuff while we are busy with hopefully something more productive.  In our view of the Cognitive Era, Watson will augment human intelligence, working side-by-side with humans to accelerate and improve our ability to act with confidence and authority. We believe that in the future, every critical decision will be informed by systems like Watson.

Innovation is not about a zero sum game – and yes, the future does need us.