For the guy who literally wrote the book on big data and its promise to revolutionize how "we live, work, and think," Kenneth Cukier's view on big data's love child, AI, is far less rosy. Citing "zillions of limitations," Cukier has poured cold water on AI's hype, ultimately concluding that he's "not sure, honestly" about the world's ability to overcome those limitations.
Not that an uncertain outcome is hampering investments in AI. Public companies talk up AI on their earnings calls (e.g., airline Air Canada highlighted its "investments in AI and enhanced airport services and enhancements" while TV and broadband company Liberty Global touted investments "in...digital tools like advanced analytics [and] AI-based applications...[because] we know [they] will ensure we're retaining customers and growing ARPU"), even as venture capitalists dump copious quantities of cash into startups: $15.2 billion in 2017 alone.
SEE: IT leader's guide to the future of artificial intelligence (Tech Pro Research)
Of that money, China can now claim the lion's share, with Chinese startups claiming 48% of all AI funding in 2017, compared to 38% for US startups, according to CB Insights data. Over 1,100 new startups emerged in 2017, with just about everyone slapping an "AI Inside" sticker on their products.
The problem, of course, is that it's far easier to market oneself as AI than to actually deliver it.
AI phone home
In the early days of big data, experts rhapsodized about the power of data to change our lives. This was still pre-AI revolution, but the arguments were largely the same as those fueling the AI hype. As Cukier explained, "Big Data enables us not to test [a] hypothesis, but to let the data speak and tell us what hypothesis is best. And in that way it completely reshapes what we call the scientific method or...how we understand and make sense of the world." Rather than probing the data for causal connections, we'd simply need to collect more data, and have that data speak for itself.
This idea that given enough data, all human insight is shallow, has since jumped the tracks of big data and invaded our thinking on AI. This technology, the most breathless AI advocates tell us, doesn't need human intervention so much as it simply needs more data.
Except, of course, that this is complete and utter nonsense.
Even Cukier has come around to the futility of just letting data have its way. Summarizing its present (and future?) limitations, Cukier says AI is "costly, hard to make work, only applies to routine perception tasks, requires a lot of training data, not 'explainable', etc." However, Cukier is also quick to call out the benefits of AI, too, like its ability to spot patterns that people can miss.
AI, in other words, is not useless. Far from it. But it's also not god, and it's not about to replace human intuition anytime soon. Or ever.
As such, if you're an enterprise desperately trying to keep up with the Fortune 500 "Joneses" with an AI project, don't be surprised when it fails the first time, or the hundredth time. Cukier's quotes on AI being "costly, hard to make work" and more are true whether you're a Mom and Pop shop or Google. No one has cracked the code on AI, so don't expect that you'll magically do so, either. Not for discrete tasks, and perhaps not for a long time.
- Special report: How to implement AI and machine learning (free PDF) (TechRepublic)
- What is AI? Everything you need to know about Artificial Intelligence (ZDNet)
- Machine learning: The smart person's guide (TechRepublic)
- Citizen AI: A business guide to raising artificial intelligence in a digital economy (ZDNet)
- How one insurance startup is using AI to slash rates for customers (TechRepublic)
Matt is currently head of the developer ecosystem at Adobe. The views expressed are his own, not those of his employer.
Matt Asay is a veteran technology columnist who has written for CNET, ReadWrite, and other tech media. Asay has also held a variety of executive roles with leading mobile and big data software companies.