Slow LLM: Slowing Down ChatGPT to Rethink Our Relationship with AI

Slow LLM: A Tool That Defies the Speed of ChatGPT

With the Slow LLM tool, ChatGPT users are encouraged to take their time. The central idea is to promote reflection and creativity by deliberately slowing down interactions. What may seem like a disadvantage actually becomes an asset for those who wish to have a more conscious control over the responses generated by the chatbot.

In the tech world, the emphasis is often on the speed of responses. Silicon Valley is the stage for an ongoing competition to provide ever faster solutions. However, the Slow LLM browser extension takes a counter approach to this trend by intentionally slowing down the process.

Unlike software that seeks to enhance the performance of language models, Slow LLM introduces a slowdown that can sometimes be perceived as annoying. It is precisely this slowdown that encourages a reconsideration of our dependence on machines and the delegation of our thinking to artificial intelligence.

How Slow LLM Works on ChatGPT

Slow LLM is not a traditional commercial product. Created by Sam Lavigne this month, it is a browser extension that imposes a delay on the responses of ChatGPT and Claude.

Installing Slow LLM is done like any other browser extension, following the instructions on the official tool website. Once activated, it directly intervenes in the internal functioning of the page by modifying the fetch function, which manages data exchanges. Thus, the responses from ChatGPT or Claude are no longer instantaneous but appear with an artificial delay, sometimes extending over several minutes.

Instead of seeing the stream of tokens scrolling in real time, the user must wait. The chatbot remains functional, but the experience becomes slower. Each response requires patience and encourages reflection before formulating a new request.

A Resistance Through Slowing Down

Why choose to slow down the responses of ChatGPT? Slow LLM addresses a deeply ingrained habit: that of copying and pasting generated responses without taking a step back. The tool positions itself as a subtle form of protest. Some fear that AI will replace our critical thinking, creativity, or problem-solving abilities.

Both students and professionals rely on AI to save time. However, this speed can sometimes hinder reflection. Slow LLM then imposes a pause on ChatGPT or Claude, encouraging users to reread, rephrase, or even question the relevance of the results obtained.

By slowing down ChatGPT, the tool forces users to think before delegating a decision to a machine. Users are confronted with their dependence and the illusion of productivity that intelligent assistants provide. This serves as a reminder that technology is not neutral and can influence our behaviors.

Slow LLM: Who Is It For?

Slow LLM is not aimed at the hurried general public. The tool targets those who are beginning to question their use of ChatGPT. The project is directed specifically at overburdened students and professionals, those who churn through prompts without really rereading the responses, always seeking to go faster.

In terms of deployment, Slow LLM remains accessible but with limitations. The extension works on Chrome locally, but users of Firefox or Safari cannot benefit from it.

However, there is another option: deployment via DNS. In this case, an entire network can be affected. A school or company can slow down ChatGPT for all its users, sometimes without them realizing it.

This approach, however, carries risks. Routing through third-party DNS exposes users to security issues. Even the creator recommends caution and personal hosting for experimentation.

So, can Slow LLM really attract users? It is still too early to say. But the idea of deliberately slowing down ChatGPT seems almost paradoxical.

The majority of users seek to save time, not lose it. Yet, this tool highlights a growing tension. Going fast, yes. But at what cost? Behind the frustration, Slow LLM reminds us that AI like ChatGPT should not replace human intervention.