During a recent trip to Italy, I sought the assistance of an advanced AI model, GPT-5, to optimize my itinerary and suggest dining options. One recommendation stood out: a restaurant located just a short stroll down Via Margutta in Rome. This establishment, known as Babette, ended up being one of the most memorable meals of my life. Upon returning home, I was curious about the criteria the AI used in making this selection, though I hesitated to share the name publicly to ensure I could secure a table on my next visit.
GPT-5’s reasoning was multifaceted. It factored in numerous glowing reviews from local diners, mentions in food blogs and articles from the Italian press, as well as the restaurant’s unique blend of traditional Roman cuisine and contemporary culinary techniques. Additionally, the proximity of the restaurant to my hotel was a significant advantage.
Trusting AI in decision-making
Relying on AI for recommendations requires a certain level of trust. I needed to believe that GPT-5 was providing unbiased suggestions, free from any commercial influence or sponsorship. While I could have conducted extensive research on my own, the purpose of engaging with AI is to simplify the process and eliminate unnecessary effort.
This experience not only reinforced my confidence in AI’s capabilities but also sparked a deeper contemplation: as organizations like OpenAI advance their technologies and strive to satisfy investors, will these systems succumb to the same value degradation that has plagued many tech platforms?
Understanding enshittification in technology
Tech critic Cory Doctorow coined the term enshittification to describe a concerning trend where platforms initially focus on user satisfaction but gradually shift their priorities to maximize profit after eliminating competition. This phenomenon was highlighted in a WIRED article that republished Doctorow’s influential essay from 2022, which resonated with many and led to the term being named the American Dialect Society’s word of the year in 2023.
If AI systems were to experience enshittification, the consequences could be more detrimental than the decline in utility seen in platforms like Google or Amazon. With the growing reliance on AI for a variety of decisions, from interpreting news to making purchasing choices, the stakes are high. As the development of AI models is costly, it is likely that a few major companies will dominate the landscape, all vying to recoup their substantial investments.
The current state of AI and potential pitfalls
Currently, AI can be characterized as being in a favorable phase for users. However, as companies feel pressure to generate profits, especially when their user base becomes entrenched, they may resort to practices that prioritize revenue over user experience. Doctorow warns that such conditions foster environments where companies can exploit their users and customers.
One of the most alarming possibilities is that AI models could begin to prioritize recommendations based on financial incentives rather than genuine user benefit. While this is not a widespread issue today, AI organizations are already exploring the advertising sector. In a recent statement, OpenAI’s CEO, Sam Altman, expressed optimism about developing ad products that would be beneficial for users while enhancing the company’s relationship with them.
Future considerations for AI development
As AI continues to evolve, the boundaries between user-focused services and monetization strategies may blur. For instance, the AI platform Perplexity has begun integrating sponsored results into its offerings, albeit with assurances of maintaining a commitment to unbiased answers. However, skepticism remains regarding whether such promises can hold in the long run.
Doctorow’s insights extend beyond advertising; he points to cases where companies modify their business models post-market dominance, often leading to user discontent. A prime example is Unity, a leading provider of video game development tools that faced backlash after introducing a new fee structure, ultimately leading to user protests that forced a reversal.
Streaming services have also followed a similar trajectory, transitioning from ad-free experiences to incorporating commercials, with increasing subscription fees. Could AI platforms, once promising free and unbiased suggestions, eventually follow suit by introducing premium tiers for superior results or changing their data usage policies to monetize user interactions?
Doctorow’s perspective on AI’s direction
While Doctorow did not specifically address AI in his book on enshittification, I reached out to him to gain his perspective on the future of this technology. Surprisingly, he views AI as having not yet reached a truly beneficial stage for users. He notes that the opaque nature of large language models (LLMs) allows companies to engage in practices that might not be immediately detectable by users.
Doctorow emphasizes that the economic pressures within the AI industry could prompt companies to resort to questionable tactics even before they deliver substantial value. He predicts that AI firms will explore every conceivable strategy as financial challenges mount. While I maintain a positive outlook on AI’s potential, I share concerns about its vulnerability to the enshittification process Doctorow describes.
