I recently came across a video from 1996 of someone making a purchase online. This example of early internet technology gave me a glimpse of what will happen with LLM technology in the future.

In the video, the user takes nearly ten minutes to buy a 35mm camera using a desktop computer. It is clunky and painful to watch. They browse product categories instead of searching. There are errors along the way. Pages take minutes to load. They find it difficult to even access the specific product page.

After watching the video, I attempted to purchase the exact same camera from my phone in 2026. It took me under 30 seconds to find and purchase the camera.

The tempo of reality in the video does not match the way the world feels today. It is safe to say advances in technology literally changed reality around us. You can feel a patience in the video that no longer exists.

A similar situation

These days, some of us are sitting around waiting on agents, with many similar problems to the person in the video. It is obvious that at some point LLM technology will get really fast. When that happens, our relationship to it will change, and culture itself will shift as a result.

When I can simply talk out an idea and have it exist in real time, I have a feeling everything will feel quite impulsive. I don't see how patience as we know it survives that reality.

I think it is time to start to consider how to design for this future. Preserving sensibilities that we value now, and maybe deciding to abandon less useful ones.

What will change

I think some of the current interfaces we rely on, will be made less relevant:

Managing multiple tasks will be reduced

Managing multiple tasks will be uncommon when there is no time to completion. Similar to how people use a web browser on their phone (one tab at a time) compared to a desktop (access to many tabs at once). You get what you want and move on.

The shift to mobile will increase

Mobile interfaces will likely become even more dominant since the desktop computer's primary advantage is multi-tasking with its windows and large screens. If the model handles context switching on your behalf, you can be removed from that part of the equation and therefore an expansive interface is less valuable.

Fluid conversation will become primary

I imagine text-based ‘prompting’ will be replaced with synchronous fluid verbal conversations. You will be communicating your primary objective, rather than specific individual tasks. You won’t need to be overly specific, because you will be able to steer and correct in real time. Voice allows this level of flexibility.

The speed of technology and the resulting future is an impossible wave to fight against, but having some understanding of what we have now will help us know what we should and shouldn’t give up.

I don’t know if people in 1996 understood how different the future would look, but I do think it makes sense to attempt to influence the future by using history to predict some of what will change soon.