One of the most provocative statements I’ve seen lately wasn’t a political comment on Twitter—it was a line from Benedict Evans’ newsletter.
“The only intelligent thing about Watson its the PR department.”
If you’ve seen the hype about Watson—whether in its crowning moment Jeopardy! or 60 Minutes story about Watson’s impact in the medical field or any of Watson’s frequent commercials—then this statement may have set you back.
But that hype doesn’t match what we’re hearing from people on the ground. I’ve talked to three people in the last six months who have tried to use Watson in their businesses, only to see projects fail miserably.
Why? Because Watson cannot do everything.
That’s because no artificial intelligence or machine learning system can.
So let’s use the hype of Watson to talk about the strengths and weaknesses of artificial intelligence and machine learning.
STRENGTH: PROCESSING BIG DATA
The amount of data that your business collects and process is growing exponentially. This means that you can’t take the same old approach to processing data.
This is where machine learning can help. By processing massive amounts of data from disparate sources quickly, a system like Watson can provide quicker trends and analysis than your traditional spreadsheet system ever could. If you can get a machine learning engine tuned correctly, it can show you things you need to know a lot more quickly than is possible in the status quo.
This is what hospitals and medical researchers are looking for from Watson. By giving doctors access to research that is on point more quickly, an AI system can provide better options for diagnosis and treatment. The 60 Minutes story mentioned earlier does a good job of touting these benefits.
WEAKNESS: TIMETABLE OF VALUE
“Like all the AI systems in use today, Watson needs to be carefully trained with example data to take on a new kind of problem. The work needed to curate and label the necessary data has been a drag on some projects using IBM’s system.” – Wired
Artificial intelligence is like a baby. When a system is first born, it knows almost nothing. Its system has basic functions like respiration and nutrition, but it can do little on its own. It needs time to learn before it can contribute to the “family” of your tech stack.
This means that the timetable from investment to ROI for an artificial intelligence product using Watson or another machine learning technology is long. It will take quite a while before your AI system is advanced enough to really deliver value to your company. So a company needs deep pockets and long horizons to really benefit from AI.
STRENGTH: EXTREME EFFICIENCY
One big advantage of artificial intelligence and machine learning is the efficiency it can provide. By processing massive amounts of data and evaluating countless decisions, an AI system can put analysis into hyper speed. The result is faster decisions that can also be better decisions.
Artificial intelligence and machine learning systems aren’t there yet—except in some specific instances—but they will get there. That’s real business value that will benefit the bottom line. If you can integrate an AI solution that builds to that point, you’ll be a hero. Just be aware that it will take time for the system to actually learn enough to make these decisions on a snap instance.
WEAKNESS: SHIFTING APPLICATIONS
AI is so new that it’s not yet proven what uses are most powerful. Instead, this technology is still in the exploration stage. That’s all well and good, but your business probably doesn’t have the time or money sitting around to be Magellan.
This is the problem with many of the Watson applications that we’ve talked to people about. They’re promised by that top-notch PR department that Watson can do anything, so they pour resources into a project. Then they find out a little too late that Watson may not be the best option for the particular use in mind.
Watson is great at some things, and it will undoubtedly be great at many more things one day. But Watson hasn’t learned to do anything and everything now. And it’s still uncertain what particular things Watson will be of most use for.
If the best uses of a technology like machine learning are still being identified, then your business needs to approach using it as R&D, not for mission-critical deployments—at least not yet.
Business is built on value, not on hype. And right now, there’s so much hype around Watson and other artificial intelligence systems that it’s hard to make sound business decisions.
After all, hype is what PR departments do.
But when you sort through the noise and embrace the strengths of AI and machine learning, you can find good business value—if not now, then in the near future.
And it could be worth a lot more than a single win on Jeopardy!