Skip to main content
SearchLoginLogin or Signup

The Future of AI in Business: Chewbacca, Chatbots, and Why I Won’t Win at Ms. Pac Man

Published onJan 27, 2022
The Future of AI in Business: Chewbacca, Chatbots, and Why I Won’t Win at Ms. Pac Man

Editor’s Note: Raiford Smith, leader of teams with global responsibility for Google’s data center planning and energy analytics, provides a general overview on the current realities and potential future of artificial intelligence (AI) in business. In this article, Smith compares Chewbacca from Star Wars to AI today, and explains the current state of AI and where the likes of SkyNet and HAL-9000 might be promoting misunderstandings.

Keywords: business AI, antifragile, ethical AI

The Empire Strikes Back is one of my favorite films. It has fantastic adventures and one of the greatest plot twists ever.1 Despite its awesomeness, I’m always drawn to the plight of Chewbacca—everyone wants him along for the ride, but only Han Solo understands him. Chewbacca’s story is familiar to anyone in artificial intelligence (AI) today—AI skills are universally coveted, but few comprehend this new tool. Yet, this won’t always be true. Writer Arthur C. Clarke stated in “Profiles of the Future: An Inquiry into the Limits of the Possible” (1962), “any sufficiently advanced technology is indistinguishable from magic.” So it will be with AI. What is ‘magic’ today will become pedestrian tomorrow, and Chewbacca’s “RRRAARRWHHGWWR” will instead be, ‘I’ve fixed the Millennium Falcon’s hyperdrive again.’

Today, human decisions can be short-sighted or error-prone, whereas fictional AIs like HAL-9000 or SkyNet are portrayed as something to be feared. In reality, AI is just another machine—an ‘intelligent’ agent that processes inputs and pursues human-determined objectives. Watching humans play AI in go or chess best illustrates AI’s capabilities—given sufficient data and an objective, AI can eventually make better decisions. Yet, AI can’t solve every problem. It is best suited to eliminating the tedium of repetitive data-rich, logic-based tasks such as operating a machine, making better forecasts, or scouring data to find better outcomes. Looking forward, AI will further reshape how customers interact with businesses beyond today’s chatbots, enabling future AI-driven customer service agents to know who you are and what you want before you chat. It will also enable better management of financial risks by spotting fraud early or identifying difficult-to-spot trends, making markets more predictable. It will also make research efforts to find the proverbial ‘needle-in-a-haystack’ faster, easier, and cheaper. AI will manage manufacturing processes, nearly eliminating quality-control issues while overseeing supply chains to avoid stockouts (even during a pandemic).

However, AI is not all-powerful and has limitations. Fears of AI-driven unemployment, misbehavior, or algorithmic bias demonstrate that AI is a work in progress. AI requires good data and struggles to predict events that don’t already exist in a data set. Yet, Nassim Taleb’s concept of antifragility (e.g., things that improve from exposure to volatility) as discussed in his book Antifragile: Things That Gain from Disorder, show AI-based processes could be designed to overcome data- or model-based limitations, offering safer problem exploration and outcomes that improve over time. But robust models aren’t enough if we can’t agree on what ‘optimal’ means. It is not clear whether ethical AI design principles will prevail over other goals, such as profit or social order. In regard to unemployment, similar concerns were raised when machines replaced manual labor, email replaced typing pools, and human-operated elevators were automated. Each time, new jobs arose to replace those eliminated and overall employment increased. The same will be true for AI.

Taken as a whole, AI has limitations but offers businesses incredible new capabilities. AI will enable more jobs, safer transportation, less-volatile markets, and high-quality, always-available goods. The adoption of techniques to manage data issues, bias, and fragile algorithm design will unlock AI’s potential even as we wrestle with defining the ‘optimal’ outcome. The journey to widespread AI adoption will no doubt have as many thrills as it has potholes, but it will also usher in the next wave of productivity and innovation—even if I can’t beat it at Ms. Pac Man.

Disclosure Statement

Raiford Smith has no financial or non-financial disclosures to share for this article.

©2022 Raiford Smith. This article is licensed under a Creative Commons Attribution (CC BY 4.0) International license, except where otherwise indicated with respect to particular material included in the article.

No comments here
Why not start the discussion?