Watson: IBM’s Huge Gamble

Rob Enderle
Slide Show

Question: Watson. Answer: What Is a Computer That Will Have Great Impact?

We know we are headed toward some kind of artificial intelligence and our most visible progress has been on Apple products with Siri. However, IBM is making a far more telling play with Watson, which is a stronger interim step, and has Watson backed up with neural networking research, which could prove to be the path to the future of truly intelligent machines. IBM has just put an additional billion dollars into the Watson project because it recognizes that this is a game changer and that the extra resources are necessary to make this effort truly fly.

Let’s talk about Watson this week.

Watson

Watson is a natural language, real-time data analysis engine. It is designed to address difficult to answer questions with ever-improving choices of complete answers. In effect, it not only provides a far better interface to data queries, it is a learning machine that will improve its accuracy through professional use. But bad training will result in decreasing reliability, much like bad information can result in uninformed people constantly in error.

This means a technology like Watson requires a unique commitment by users who know the subject matter to train the system so that other users can most benefit. This is not a usual requirement. Typically, a system is installed and users are trained in its use. With Watson, the machine actually will need more time to reach its potential because it needs to be trained to apply the knowledge it contains intelligently.

Now, once this is done, someone with little expertise can be trained to use the system and approach problems as if they were experts, but experts are needed to do that work. Since areas of specialization tend to change rapidly, Watson will likely need constant oversight by experts to assure the quality of the services it provides to those less expert. Used properly, Watson not only should improve the speed and accuracy of expert staff in the area where Watson is deployed, but massively expand this expertise to everyone who uses the system. In effect, it is an expert enhancer and multiplier, allowing you to do with a small pool of experts what would otherwise require far more.

This brings up three potential problems that need to be considered.


Three Problems

Skills are tied to status. Experts often relish this status and don’t want to share it with others. They also don’t like to be seen as wrong. Watson tends to level the field so that experts don’t seem that special, and once trained, it is likely to be more accurate more often. Experts don’t like to be wrong. This means that the experts you need to assure that Watson reaches its potential may not be all that helpful in the process because they may see the system as a threat. Care will need to be taken to make sure those feelings don’t have an adverse impact on the effort.

Second, a lot of folks that present themselves as experts aren’t. If they are involved the process of training Watson, they’ll likely corrupt the process on purpose or accidentally. This means you’ll need to assure that the people training Watson actually have the skills needed for the project. That means both background checks and oversight to catch those that can fool background checks.

 

Finally, executives often like to make decisions with their gut and drive their decision with their title. If used properly, Watson should identify and highlight many, if not most, related bad decisions. Care will need to be taken so that the executive isn’t embarrassed by the outcome, otherwise, he or she may be likely to work to pull funding. This also means that helping executives make the right decision and assuring they get credit for it will go a long way toward assuring the continued funding of the project.

Wrapping Up: Huge Benefits with Watson

In the end, Watson, properly trained, implemented and used could massively reduce the bad decisions a company makes, improve the good ones both in number and quality, and increase the firm’s agility. It still requires, however, that someone ask the right questions in a timely manner, and that Watson not be screwed up as a result of bad staffing or bad data. If you are going to implement Watson, you really need to step back first and realize that this is something very different. It is also something far more powerful than you have likely ever seen. This one system has the power to restore IBM to its full former glory, but it requires that a lot of people behave differently toward it than any other technology product. And that is IBM’s big gamble.



Add Comment      Leave a comment on this blog post

Jan 16, 2014 7:35 PM Francis Deumart Francis Deumart  says:
You really shouldn't write articles if you have no idea about the subject. As just one example, Watson is not based on 'neural network research' any more than the latest Corvette is based on 'carburetor technology'. As another, teaching expert systems has been a field of study for 20+ years. It is not a new concern that will take anyone by surprise. IBM may or may not return to their 'former glory' but it's always worth pointing out that they sit at #20 on the Fortune 500, around the same level as Chase and with double the revenue of Google. Not too shabby. Reply

Post a comment

 

 

 

 


(Maximum characters: 1200). You have 1200 characters left.

 

 

Subscribe to our Newsletters

Sign up now and get the best business technology insights direct to your inbox.


 

Resource centers

Business Intelligence

Business performance information for strategic and operational decision-making

SOA

SOA uses interoperable services grouped around business processes to ease data integration

Data Warehousing

Data warehousing helps companies make sense of their operational data