A.I. can cure click fraud’s $100 million problem, and it won’t steal your job

IBM Watson isn’t just about beating humans at Jeopardy or recognizing English conversational speech with a meager 6.9 percent word error rate. It is part of a new breed of A.I. solutions that could change the way we do business forever.

Fresh from the stage at NASDAQ Marketsite — as part of AdWeek New York — Steve Gold, CMO at IBM Watson, spoke with me about the future of A.I., machine-to-human interaction, and advertising technology.

The first question most people ask about A.I. is, “Will it replace humans?”

“What’s interesting about job displacement, not only in A.I. but in technology in general, is it’s the same conversation we’ve been having since the dawn of the computing age,” Gold told me. “IBM mainframes created jobs, rather than removing them.”

And while some jobs will be replaced, just as they have been throughout the course of time, new roles appear every year. What is more interesting is whether A.I. can elevate existing positions and make us all better at what we do.

“Let’s take consumer interactions as an example,” Gold said. “Only half of inbound calls are resolved within the first contact. Why? Because there is variability in the quality of the agent, the training they received, the breadth of their knowledge, and their understanding of the problem. Virtual agents can improve the process by qualifying the incoming call — or chat — before handing it over to exactly the right agent at the right time.”

In other words, we can train A.I. to look after the part of the process that is repetitive, enabling support agents to focus on what is important.

So A.I. becomes a tool to enhance our careers.

“It’s really about working out how we extend the individual,” Gold said. “How do we make you a better journalist?”

That’s a question many have asked.

But what happens when the machines start talking to the machines, and what if they start to lie to each other to get the best deal for their owners? Are we facing an accelerated race to the bottom in every aspect of business?

“When it comes to machine-to-machine interaction. A.I. isn’t the natural accelerant here,” Gold said. “A.I. is much more aligned with human-to-machine interaction, firstly, but there is also an inherent issue with attempting malevolence — machine learning and big data. These systems live without bias — they’re not taught to be immoral. They’re taught based on outcomes and new information.”

Gold then gives an example that might pique the interest of rapper B.o.B.

“I could teach A.I. that the world is flat,” Gold said. “But A.I. can infer from other statements and data that this learning just isn’t right. It’s like influencing a child early on in life. As they’re schooled, they overcome those learnings and gain a better understanding of what is right through data, peers, and their environment.”

So with A.I.’s ability to understand what is true and what isn’t and to discover patterns, could it be used as a potential answer to the problem of click fraud, something that is costing app marketers alone $100 million this year?

“The ability to discover patterns is important here,” Gold said. “By determining what click fraud looks like, and understanding those trends, we can fix the problem.”

Advertising is already being enhanced by IBM Watson via its new Ads solution, which offers interactive advertisements that allow the consumer to ask questions. As A.I. becomes more prevalent in adtech, we’ll start to see ads that look less like a brash interruption and more like an assistant, providing us with just what we need, when we need it, and with the ability to answer our queries directly.

It’s a brave new world, and IBM Watson is helping write an exciting chapter in cognitive computing.