I have always liked Roger Federer, the ultimate gentleman sports hero. If at all I were to be a sportsperson — for which, let’s face it, my best hope is a favorable reincarnation in the next life — I would like to be Federer. His mastery of the sport has been matched only by his genuineness, curiosity, and class off the court. So, I wasn’t too surprised when Federer was trending the other day for his great commencement speech at Dartmouth. Par for the course, as we aspiring sportspeople like to say.
Anyway, his statement that he won only 54% of his points, despite winning almost 80% of his 1,526 singles tennis matches, got me thinking. The 4 percent above the average must have included the most impactful points in his matches. It reminded me of a well-known paradox of technology: efficiency does not explain success, success explains efficiency. It is this paradox that I’d like to focus on, to share a few thoughts on how some Artificial Intelligence wins may be more important than others.
The Ten Paradoxes of Technology
The statement that efficiency does not explain success, success explains efficiency, is one of the ten paradoxes of technology published by Andrew Feenberg. Feenberg is an American philosopher at the School of Communication at Simon Fraser University in Vancouver. His work on the philosophy of technology has been widely praised as positive and holistic. His work on the ten paradoxes of technology is particularly timely and relevant.
Feenberg published The Ten Paradoxes of Technology in 2010 with a bold assertion that many of our ideas about technology are false. Our fault is founded on the assumption that technology is a tool that society can choose to use or not. According to Feenberg, there is no need to make such an “either-or” choice. There are as many possible technological options as there are paths toward progress. This can certainly seem like a weighty pronouncement, one that warrants several beers with a close circle of friends to dissect effectively. But my zero-calorie version runs along these lines: society and technology develop together. Society shapes technology, and technology shapes society.
I expect to write more about these ten paradoxes in the future, but here are a couple of fascinating appetizers for now:
- Paradox # 1 – Paradox of the parts and the whole: Technology does not have meaning without a broader context. Consider the question: do birds fly because they have wings, or have wings because they fly?
- Paradox # 6 – The means are the end: The technologies we own symbolize who we are, and our social status can be determined by the technologies we use.
There’s a lot more where that came from, but let’s get back to Federer and AI.
AI Perspective: Efficiency Does Not Explain Success, Success Explains Efficiency
The efficiency vs. success paradox is number four of the ten. This law is also known as the “paradox of the frame.” The efficiency of a technology does not explain its success, but its success is the reason for its efficiency — we should study the context under which it was successful to understand its reasons. So, the fact that Federer won 54% of points doesn’t explain why he won 80% of matches. Hence, the context that Federer hints at in the speech needs to be studied. No, it is not beer pong, which he hilariously mentions eight times; it’s defining talent very broadly, e.g., grit, discipline, trusting yourself, focusing on winning one point at a time, etc.
I believe there is merit in applying the fourth paradox of technology to AI tools. So, here it is… My framing of the fourth paradox of technology as applied to AI tools is as follows:
There are so many good (read: efficient) AI tools, but few will spell success for you. Your success will determine which of the AI tools is “good.”
But How Do I Decide Which AI Tools to Use?
Before you say, “Thanks for nothing, Tony,” hear me out. I am not suggesting that we wait until the last chapter of your AI tool investment story to find out if the tool was good. That’s what makes it paradoxical — a paradox is essentially an absurd or self-contradictory statement that happens to express a possible truth. My point is that, for powerful, evolving technologies like AI, we need to ask different questions to evaluate what tools to use. Not “Is this tool good,” but “How do I know I will be successful given the holistic context.” This frames the broader social, environmental, and contextual challenges in addition to the quality of the AI tool itself.
Tips for how to go about selecting the most suitable AI tools:
- Experiment with many, choose a few: If you don’t try several — in a quick, iterative manner — you’re missing out on possibilities.
- Test AI tools in a holistic context: You require feedback on not just the tool’s performance but also the social and contextual challenges. This becomes even more important if we want to use AI for decision-making.
- Look for proven use cases: Proven success stories in similar contexts are worth more than technical superiority.
If these do not work for you, there’s always Federer’s beer pong. But perhaps not. I would be more inclined to bet on his winning life strategies than on his jokes.
Tony Saldanha is a News Columnist at Grit Daily. He is the President of Transformant, a consulting firm specializing in assisting organizations through digital transformations. During his twenty-seven-year career at Procter & Gamble, he ran both operations and digital transformation for P&G’s famed global business services and IT organization in every region of the world, ending up as Vice President of Global Business services, next Generation services. He is an advisor to boards and CEOs on digital transformation, a sought-after speaker, and a globally awarded industry thought leader.