The Chatbot Tradeoff: Part 1
Technology and People
“To ignore technology is a decision—one that turns us into a cog in someone else’s machine. . . . Why would any of us accept such a passive role in our own lives when we can hold those who work with technology to account so that they work for and with us, rather than against us?” — Stephanie Hare in her book Technology is Not Neutral
As major companies rapidly roll out generative AI technology, many of us feel like we are entirely at the mercy of those companies. We feel forced to adapt. Will citizens and businesses have any say in what will certainly impact our kids’ educations, our jobs, and even the global economy? As Stephanie Hare asks, “Why would any of accept such a passive role . . .?”
Well, we might not be able to stop large corporations like Microsoft and Google from offering chatbots and digital image generators, but no one is forcing us to use them. We are, after all, still consumers endowed with free will. We can choose what we buy and use. If most individuals and businesses choose to not use these AI programs, the companies would not profit from them.
So, what if we approach the question of using generative AI as a basic consumer transaction? A starting point with any purchase is to evaluate the pros and cons of “buying in.” The consumer’s first question is simple: What is the primary benefit of generative AI programs? These technologies can be used in numerous ways, of course, but all the benefits can be boiled down to one common denominator: efficiency.
Efficiency involves factors such as time, productivity, and revenue. For millennia, people have sought ways to do more in less time and/or with less effort, but Americans dramatically elevated the cultural value of efficiency in the early 1900s, at the height of the Industrial Revolution.
Frederick Winslow Taylor led the way. He became known as the “high priest” of the “efficiency movement,” which used the mantra “time is money.” He promoted a method of reducing a production process down to small components, or into individual standardized tasks (James Suzman, Work, Penguin Press, 2020, p. 334).
Factory owners, of course, loved Taylor’s efficiency movement because it reduced labor costs while also increasing productivity and profits. After Taylor died in 1915, a large crop of new “efficiency experts” emerged, people who worked evangelistically to reduce labor costs while amplifying the profits of companies and lowering costs for consumers.
The efficiency movement became so popular that it even spread to housework. In 1950, co-authors Mary and Russel Wright published their Guide to Easier Living for housewives who spent all day alone while the kids were at school and the husbands were practicing Taylorism in the office. Women applied Taylor’s methods to cooking, shopping, dishwashing, and laundry—with the hope of finding more time for enjoyable activities. Some housewives came to think of themselves as household “production engineers.”
As increasing numbers of companies began applying Taylorism to improve productivity and profits, free-market competition put competitive pressure on company leaders to adopt efficiency protocols. The most efficient companies would, at least theoretically, be able to dominate less-efficient firms. An “arms race” emerged, with companies perpetually seeking new technologies designed to do work faster and cheaper.
Efficiency gradually became a quasi-religious value in US culture—often at the expense of quality, aesthetics, creativity, and human flourishing. Products, such as houses and cars, and clothing, became mass-produced and therefore boring and homogenous. (The film Ford vs. Ferrari illustrates this trend in the US auto industry.) And then the efficiency movement spread globally.
“With the world’s economy resting more and more on competition between manufacturing enterprises, someone had to notice that the key variable in the arithmetic of production was always time,” writes James Gleick in his classic 1999 book Faster: The Acceleration of Just about Everything. “Taylorism is the ideal of efficiency applied to production as a scientific method—humans and machines working together, at maximum speed, with clockwork rationality” (p. 213).
Generative AI is essentially the application of Taylorism to the work of writing, visual and graphic arts, filmmaking, and music. Other types of artificial intelligence, such as programs that help with shipping logistics and pharmaceutical development, are also tools for reducing the time and cost of production.
AI might be a new efficiency tool, but it is fueled by the same value system—the old efficiency movement that emerged during the Industrial Revolution. And, as occurred in the past, the religion of efficiency will continue to impose on society a tension between productivity and human well-being.
Taylor’s biographer Robert Kanigel stated the tension of efficiency this way: “Each day we reap the material benefits of the cult of workplace efficiency that [Taylor] championed, yet we chafe—we scream, we howl, we protest—at the psychic chains in which it grips us” (as cited by Gleick, p. 215).
The Costs of AI
As rational, free consumers, we should next consider the negative costs of generative artificial intelligence before we buy in. We are not required to use these tools any more than we are required to eat at Taco Bell. We should ask whether the benefits of AI efficiency outweigh the costs. Why buy something that imposes more costs than benefits?
To keep this article from getting too long, we will address just one negative cost of generative AI. In subsequent articles, we will consider other serious harmful effects of these technologies.
The Disregard for Truth
Chatbots, the ones that generate written responses to human prompts, regurgitate data scraped from the internet ocean. That ocean of data is “dirty,” meaning that it is polluted with massive quantities of false information. Because these programs cannot discern between truth and falsehood. Therefore, there is a significant likelihood that chatbot responses will be polluted with misinformation and even disinformation. Thus, a consumer needs to decide whether efficiency (time savings) is more valuable than truth. Do we want a product that spreads false information? Is that a good consumer tradeoff?
Discerning what is true today is already a serious problem, of course, but chatbots will likely make matters even worse. There are at least two reasons for this.
First, today’s traditional search engines provide users with a long list of websites. That list is undoubtedly full of shoddy information, but the user can see the sources of information and evaluate whether they are reputable and reliable. For example, we can choose to only read the content published by organizations that are economically and legally held accountable for the information they disseminate.
By contrast, Microsoft Bing’s chatbot, for example, removes the list of websites and offers a summary text without citing any sources. Thus, the consumer is asked to trust the chatbot’s information without allowing him or her to know the origin of the information. In short, the user is asked to place blind faith in the chatbot. If consumers are sincerely interested in truth, as we should be, then we will have to spend extensive time fact-checking the information through other means. The time spent doing that could result in zero net efficiency gains.
Second, generative AI programs—text, audio, images, video—make it extremely easy and efficient for nefarious actors to intentionally spread false information throughout the entire information ecosystem. These individuals use AI tools to disseminate propaganda designed to sway public opinion.
“The first and most obvious threat is that AI-enhanced social media will wash ever-larger torrents of garbage into our public conversation,” write Jonathan Haidt and Eric Schmidt in The Atlantic. “In the age of social media . . . propaganda doesn’t have to convince people in order to be effective; the point is to overwhelm the citizenry with interesting content that will keep them disoriented, distrustful, and angry” (https://www.theatlantic.com/technology/archive/2023/05/generative-ai-social-media-integration-dangers-disinformation-addiction/673940/).
The outcome of all this false information and disinformation will be the further erosion of trust. We as a society are already grappling with this problem, struggling to know who is telling the truth. Conspiracy theories run rampant partly because social media platforms are not held legally accountable for what they publish. Generative AI will make matters much worse.
To the extent that society loses its ability to trust businesses, government, or each other, it will lose its ability to function. That is the opposite of efficient! Ironically, as the efficiency movement drives the widespread use of generative AI programs, those programs have the potential to reduce the healthy function of social relations and the economy—a net reduction in economic efficiency and human flourishing.