The Overhype of Generative AI: A Q&A with Dr. Eric Siegel

We're bombarded with visions of super-intelligent AI taking over the world, but is that realistic?

In this interview, Dr. Eric Siegel, former Columbia professor, leading ML consultant, and author of The AI Playbook: Mastering the Rare Art of Machine Learning Deployment, argues that predictive AI—aka enterprise machine learning—offers concrete value that generative AI has yet to prove.

We'll explore why many ML projects fail and how to bridge the gap between business and data teams. Discover how businesses harness ML to tackle real-world challenges and gain a competitive edge.

Why do people mistakenly believe AI will become as capable as humans – or potentially pose a deadly risk to the human race?

The wish fulfillment promised by artificial general intelligence (AGI) – software capable of any intellectual task humans can do – is so seductive that it's nearly irresistible. By creating the ultimate power, we achieve the ultimate ego satisfaction as scientists and futurists.

By building a system that sets its own goals and autonomously pursues them as effectively as a person, we externalize our proactive volition, transplanting it into a new best friend for humankind for whom we hold the very highest regard and with whom we can potentially empathize. By creating a new life form, we realize every last bit of the as-yet-unrealized potential of our general-purpose machines known as computers. By recreating ourselves, we gain immortality.

By creating a single solution to all problems, we transcend any measure of financial reward to gain infinite wealth. ML thought leader and executive Richard Heimann calls this the single solution fallacy. Rather than solving the world's many problems one at a time, we solve them all in one fell swoop with the ultimate silver bullet.

We need not fret about global issues such as climate change, political instability, poverty, or health crises. Instead, once an artificial human comes into existence, it will continue to advance itself to become as capable a problem-solver as the human race could ever be.

What is a good strategy to bridge the divide separating business professionals and data professionals, which routinely prevents ML from succeeding?

Here’s the problem. ML is the world’s most powerful generally applicable technology. However, ML can only improve large-scale operations by changing them. Therefore, an ML project shouldn’t be viewed as “a technology project.” Instead, to make an impact, it must be reframed as a business project meant to improve operational performance, with ML as only one component—one that’s necessary but not sufficient.

With the attention overwhelmingly focused on the technical portion and its execution, the industry has failed to establish a widely adopted business practice for executing the whole other half of a successful ML project. As a result, new ML initiatives routinely fail to deploy.

My solution to this is bizML, a six-step discipline for running an ML project so that it successfully deploys. You can read more about this in my latest book.

What is your explanation for why most enterprise ML projects fail to deploy while a select few succeed wildly?

The root cause of most ML project failures is a lack of rigorous planning for deployment – planning for the operational change that integrating a predictive model would enact. Since the world conceives of ML projects as technical projects involving the most advanced number crunching, it is assumed the technical project will deliver value. This is a grave misconception.

The value is only captured by *changing* – and thereby improving – large-scale operations, as guided by the predictions provided by an ML model.

Do you have a contrarian assessment of the overzealous hype surrounding artificial intelligence and why hardly anyone asks – or measures – how good AI technology is?

These are two different items. As for the hype, the popular narrative is that we are headed toward AGI. This is a myth. It is the novel Mary Shelly would have written if she had known about algorithms.

As for why we aren't measuring how *good* AI is – how well it performs quantitatively and how much business value it could deliver, depending on how it's deployed – the focus is misguided.

Author's Tip

“When evaluating ML models, data scientists focus almost entirely on technical metrics like precision, recall, and lift, a kind of predictive multiplier (in other words, how many times better than guessing does the model predict?). But these metrics are critically insufficient. They tell us the relative performance of a predictive model — in comparison to a baseline such as random guessing — but provide no direct reading on the absolute business value of a model. Even the most common, go-to metric, accuracy, falls into this category. (Also, it’s usually impertinent and often misleading.)

 

Instead, the focus should be on business metrics — such as revenue, profit, savings, and number of customers acquired. These straightforward, salient metrics gauge the fundamental notions of success. They relate directly to business objectives and reveal the true value of the imperfect predictions ML delivers. They’re core to building a much-needed bridge between business and data science teams.

 

Unfortunately, data scientists routinely omit business metrics from reports and discussions, despite their importance. Instead, technical metrics dominate the ML practice — both in terms of technical execution and in reporting results to stakeholders.”

MIT Sloan Management Review

How are organizations actively using ML to supercharge operations and create competitive advantage across a number of industries?

ML innovates in a straightforward, albeit disruptive way. Don’t let the glare emanating from this glitzy technology obscure the simplicity of its fundamental duty: For most business applications, the purpose of ML is to issue actionable predictions—which is why it’s also sometimes called predictive analytics or predictive AI.

Although learning from data in order to generate a predictive model deserves as much “gee-whiz” admiration as any other feat of science or engineering, that capability translates into tangible value in an uncomplicated manner: The model generates predictive scores, which in turn drive millions of operational decisions.

In this way, ML combats our most significant risks—including wildfires, climate change, pandemics, and child abuse. It boosts sales, cuts costs, prevents fraud, streamlines manufacturing, and strengthens healthcare.

For more insightful interviews, join The CTO Club's newsletter.