Innovation

CB Insights AI tool predicts next big thing in tech

News media can be remarkably accurate in capturing popular demand for new technology. Now, a new tool from CB Insights can help identify which trends are worth paying attention to, and which aren't.

Image: iStockphoto.com/Highwaystarz-Photography, Getty Images/iStockphoto

Is 2016 the year of AI? Big data? Virtual Reality? How do we know?

The answer, said CB Insights, may lie in media coverage. Whereas traditionally, many predictors relied on tracking investments in different areas or talking to specialists to uncover hot areas in tech, CB Insights' new tool called Trends now analyzes media reporting on different topics to uncover the next big thing.

According to CB Insights, "When aggregated and analyzed, news media offers a powerful real-time method to discover, predict and plot the arc of rising expectations and excitement (and potential eventual disappointment) for emerging technologies."

Compared to analysts, who may vary widely on predictions (on the drone market, for example, experts have cited market value from $1.27 all the way to $10 billion by 2020) media trends seem to be a more accurate way to gauge enthusiasm in an area.

The new tool works via CB Insights' SaaS platform. Clients can compare trends with other details about companies, like "funding or business momentum" to see what to pay attention to.

TechRepublic contacted a spokesperson at CB Insights to find out more about how this works, how it can outperform analysts, and why it's important.

What has the reaction been to Trends so far from clients using the platform?

Very favorable, with some surprises. We're seeing clients use Trends in their work including public blog posts. Here is an example of an analysis by a VC at Technicolor Ventures.

In terms of surprises, we've seen clients use Trends to dig into certain topics more than others. As an example, chatbots and bots have been much more popular than bigger areas such as big data.

What kind of data is CB Insights looking at?

For Trends, we only look at news media—everything from blogs to newspapers. Each piece of information has to be treated differently. For example, we do a lot of work around sentiment. Discerning sentiment in an article is very different than a tweet and so each type of information input must be treated in a specific way as generalized frameworks yield bad, imprecise results in our experience.

SEE: Sentiment analysis: Understanding customers who don't mean what they say (TechRepublic)

The uniqueness of what we do lies in the technology we've built to classify and extract data. As an example, partnership articles are different than financing articles or M&A media mentions. What we've found is that generalized classification and entity extraction rules don't work. As a result, we've built very specific machine learning to manage the process of crawling, classifying, and extracting data. We call this technology The Cruncher.

What do you mean by the "Pundit Industrial Complex?" How does CB Insights offer something unique?

The Pundit Industrial Complex (PIC) is the army of consultants and industry analysts who make up predictions about where technology and tech trends are going.

In baseball and political forecasting, data and probability have been used to make better predictions. In baseball, there was Moneyball or sabermetrics. In political forecasting, Nate Silver and 538 showed this.

SEE: Report: The top tech trends impacting the enterprise (TechRepublic)

But, when it comes to technology trend forecasting, the process is very antiquated and void of any technology.

Our technology-driven approach is unique. Instead of relying on pundits, we rely on probability and software to analyze millions of inputs on media, venture capital, sentiment, patents, and more to predict technology trends.

What have been the biggest victories of CB Insights? What trends have you caught before anyone else?

We've been early to some trends including Insurance Tech and AI. On the flip side, our data saw the challenges of food delivery and subscription e-commerce early.

Do you worry that relying on media for "trends" might just reinforce the hype around certain ideas? That the excitement will fade away?

Some things are definitely fads, but that doesn't diminish the value of the media in highlighting and seeing trends.

What we've noticed is that trends follow a Media Attention Path (MAP). The early part of any MAP is the discovery phase. And in the discovery phase, the trends we're finding are not mainstream and hence are covered by more niche blogs or industry-specific publications.

Sometimes, those trends don't break out and just fade away, or stay niche. But sometimes, those trends that are discovered by niche publications start to gain greater adoption and get covered in more mainstream publications and start to become more well-known.

What kinds of businesses will be most interested in Trends?

Our primary clients are corporations who are trying to understand where technology is going. It will be corporate innovation, strategy, venture, and product groups as well as the CIO who are the most frequent clients.

They're using it to predict the next market they should enter, how to position their products, what startup companies they should be worried about, or to see who they might want to acquire or invest in.

What, specifically, in terms of trends, do you see for the future of AI? For VR? For autonomous cars?

For AI, right now, we're seeing a ton of activity and chatter about chatbots and bots. We're also seeing lots of industry-specific applications, i.e. AI for customer support, for shopping, for financial services.

For VR, we've noticed that two years ago, the conversation was all about Oculus and that now, the competitive dynamics have shifted and the market has a much more diverse slate of players. In other words, Oculus is no longer the only game in town.

VR also clearly trumps AR in discussion/popularity.

Per Trends, autonomous vehicles also has become less synonymous with Google as competitive intensity in the space has increased. Interestingly, GM is doing more in the space while Apple's efforts remain a bit rudderless.

How exactly are you assessing media coverage in your Trends report? How would an article from TechRepublic on AI be woven into the research, for example?

We'd capture the article in TechRepublic and so, if someone looked up AI on Trends, the article would be included in the summary trend that is provided.

What is more interesting, however, is if someone says "artificial intelligence and banking," that would reveal the TechRepublic article only if it talks about AI as it relates to banking.

The value, ultimately, is in seeing if a lot of folks are talking about AI and banking and if that trend is increasing or decreasing over time.

Also see...

About Hope Reese

Hope Reese is a Staff Writer for TechRepublic. She covers the intersection of technology and society, examining the people and ideas that transform how we live today.

Editor's Picks

Free Newsletters, In your Inbox