AI: Potential Promise or Just Another Technological Trend?

According to Keir Starmer’s “AI opportunities action plan,” which was launched back in January, AI will be infused “into the veins of this enterprising nation.” Whether intentional or not, such vivid, even ecstatic vision raises an important question: is artificial intelligence a game-changer or just another exaggerated tech gimmick?

The authors of a recent book that challenges the quick development of AI contend that we are being duped. They contend that rather than being revolutionary, AI is just a “bill of goods” being sold by a few tech behemoths hoping to profit from human work, data, and creativity in exchange for poor automated alternatives.

Think about large language models such as ChatGPT. They function similarly to clever auto-complete software, producing words that may sound intelligent but may not actually be understood. The fact that they are trained on a vast variety of datasets, many of which contain copyrighted content—sometimes without authorization—exacerbates this problem even more. The effects are already being felt: according to a Society of Authors poll, more than 25% of creatives have lost their jobs as a result of generative AI.

In their controversial usage of the term “synthetic text-extruding machines,” the authors Bender and Hanna compare LLMs to industrial enterprises that transform useless raw materials into things that appear to be useful. Since they contend that AI-generated music and visuals are no different from widespread copying, this argument can also be used to various types of media. In fact, OpenAI is being sued by the New York Times for these problems.

The ramifications are not limited to the arts. Our ability to comprehend material in its totality is being stolen by AI-generated summaries, which are taking over search results pages. We overlook the subtleties and background information that come with citing several sources. The authors caution that although some people praise AI for its effectiveness, it is usually used as a tool to replace human labor. One example is when, within days of the employees’ vote to organize, a chatbot replaced the human personnel at an American eating disorders helpline.

However, not every AI application is negative. The authors do bring forth “reasonable use-cases” such as physician transcription software and medical image processing. Additional beneficial uses include improving industrial operations, optimizing energy grids, and even expanding the frontiers of medical research, such as DeepMind’s Nobel Prize-winning protein folding work.

Finally, Bender and Hanna serve as a reminder to exercise critical thought. Something should not always be done just because AI is capable of doing it. We need to examine the morality, prejudice, and actual consequences of entrusting decision-making to unaccountable machines.

They issue a straightforward warning: we may become disoriented if AI becomes “mainlined” into society.