The question of “man versus machine” has fascinated people for centuries. Many works of literature across cultures and history involve competition between humans and gadgets, from folk tales such as John Henry to movies such as The Terminator. Artificial intelligence (AI) is the latest technology said to compete with humans, but on a new frontier. While these stories feature physical conflicts, only recently has technology challenged humanity intellectually. Below are four examples of AI competing with human experts at their mind games of choice, and each contest reveals the unique mental virtues of humans that must be embraced by professionals in the era of ChatGPT.
One of the most notable competitions between AI and humans was when IBM programmed Deep Blue, a supercomputer designed to challenge the world’s then-top-ranked chess player Garry Kasparov. Kasparov is one of chess’s all-time greats, having reigned as the world number-one from 1984 through 2005, the longest streak ever recorded. Deep Blue could analyze over 200 million positions per second and had a memory of moves and counters from decades of games. This greatly advantaged Deep Blue, as chess’ only variables are each player’s choices and which player moves first. Deep Blue’s ham-fisted approach of evaluating millions of possible moves and recalling their counters failed against Kasparov’s unpredictable strategy at first, as Kasparov won their match four games to two in 1996, despite Deep Blue being reprogrammed between games. A rematch was held in 1997, with the Deep Blue team winning by 3.5-2.5 games. To paraphrase Neil Armstrong, this was “one giant leap for machine-kind,” as many expected a human grandmaster to defeat a computer. Furthermore, Deep Blue revealed that chess can be conquered by brute force, as the fixed board size and number of pieces limit each player’s options.
IBM designed another supercomputer to challenge humanity in the quiz show Jeopardy! In 2011, IBM pitted Watson against human champions Ken Jennings and Brad Rutter, with Jennings holding Jeopardy!’s longest winning streak. Jeopardy! was a new and unpredictable challenge for IBM, as contestants must respond to a prompt selected from one of many categories and phrase their response as a question. To win, Watson must interpret each prompt live and recall a correct answer. The IBM team had nearly perfected Watson’s ability to do this, as it cruised to victory aided by its database of four terabytes containing over 200 million pages of information. Although it sometimes faltered (Watson answered “What is Toronto?????” in response to a Final Jeopardy question where the category was “U.S. Cities”, surprising Jeopardy!’s Canadian host Alex Trebek), Watson demonstrated advances in AI’s ability to immediately answer complex and indirectly worded questions without searching the Internet.
Since the invention of Deep Blue, increasingly powerful new chess engines had mastered the centuries-old game. Observers then moved the goalposts, deeming the ancient Chinese board game of Go to be one where the human touch of an expert was unbeatable by a computer. Unlike chess, the possible moves in Go are nearly limitless, and the player who captures more of the game board is the winner. Go’s unpredictability thus required a new approach. Google’s DeepMind team designed AlphaGo, a Go-playing program that uses machine learning to plan moves while seeing the whole board at once, allowing it to independently adjust its strategy during and between games, unlike the manual adjustments required by Deep Blue. In 2016, AlphaGo challenged Lee Sedol, a top-ranked professional player. AlphaGo defeated Sedol four games to one, using its ability to monitor the entire board simultaneously, free from the limits of the human body that could cause even a master player to make mistakes. AlphaGo made unconventional choices during the match that confused observers but were key to its victories. Sedol was impressed by AlphaGo’s performance, and players learned new strategies from AlphaGo that brought innovation to the game.
It may be hard to imagine AI having flaws after studying the three previous examples of software beating humans at their own game. However, the forum of debate exposes AI’s limits. Unlike chess, Jeopardy!, or Go, the subjective judgment of an audience or a panel of judges measures debate performance. An AI must appeal to the hearts and minds of human observers to “win” a debate. In 2019 IBM released Project Debater, an AI machine said to be the first computer capable of debating a human. Its challenger was Harish Natarajan, a master debater and grand finalist in 2016’s World Debating Championships. Project Debater made a powerful opening statement, combining effective logical and emotional appeals, and finished with a strong closing argument. However, it struggled during the rebuttal portion. Despite its large database, Project Debater failed to counter Natarajan’s arguments, as it restated its original arguments with little acknowledgment of Natarajan’s points, likely because Project Debater was not permitted to obtain new information from the Internet. Natarajan was declared the victor based on audience poll results. Although Project Debater’s conversational skills were a precursor to ChatGPT’s ability to answer an incredible number of questions, its weaknesses show the current limits of AI, notably how it lags behind humans in continuing conversations and responding to new information independently.
These four examples reveal many insights about the progress of AI’s ability to challenge humans intellectually, where its strengths and weaknesses lie currently, and highlight qualities that are unique to the human mind. AI has already surpassed a human’s ability to store and recite information. In the workplace, the “Encyclopedia Brown” knowledge worker whose strength is memorizing information will become less valuable as access to software that can provide information rapidly becomes widespread. Attention to detail, often deemed crucial to professional success, is another area where AI excels, as programs like AlphaGo “see” the areas they are monitoring at once while avoiding oversights that even well-trained humans may make. If all someone has to offer is precision and memorized knowledge, they provide little that an AI cannot. This goes double for the stereotypical “eccentric genius” who provides a wealth of expertise at the price of a toxic personality that reduces morale and prevents collaboration.
On the contrary, the cases where AI and humans have gone head-to-head reveal traits seen in humans that are rarely found in a computer program. First, self-awareness is absent in every AI but is (hopefully!) found in many people. In marketing, self-awareness involves understanding why an individual does what they do, feels how they feel, or thinks how they think. An AI will execute tasks without an internal understanding of the processes behind its actions. In each of its contests against human minds, AI has made revolutionary insights despite not knowing whether its choices are innovative or foolish. “Starting with why” has become a maxim in the corporate world, and it is vital to understand how we think, how AI “thinks”, and to evaluate our thought processes and principles to create better solutions. Another uniquely human quality revealed in the contest with Project Debater is the ability to connect with other human beings. Professionals who relate to clients and colleagues, understand their concerns, and present perspectives and solutions that resonate with them offer skills not easily mimicked with an algorithm. Finally, a human will be a far better storyteller than any chatbot. Marketing is essentially storytelling, and a marketer who can distill a brand’s story into a message that lands with the target audience will be a rainmaker. While AI is a tool that will augment our knowledge, it will not provide more value than a marketer with the previously mentioned skills for quite some time.
I will conclude with a perspective from none other than ChatGPT:
“Marketing and advertising professionals possess creativity, industry-specific knowledge, and experience that cannot be easily replicated by AI. They excel at crafting messages, have insights into consumer behavior and market trends, and often work in teams, providing personalized attention and building relationships with clients. While AI can support and enhance their work, it cannot replace the human touch that is essential to the industry.”