10,000 Hours and AI
Over the past few weeks I’ve had a number of conversations with people about tools like ChatGPT and DeepSeek and what they mean to overall creative endeavours. A lot of people, myself included, are under the impression that if the value of human creativity is effectively zero, fewer people will be encouraged to invest the time needed to become proficient in a skill. Why struggle for 10,000 hours1 when a computer can give you something that is technically perfect within a matter of seconds, after all? However, there does seem to be a trend developing where people are beginning to differentiate when to use AI and when to do the work to create something new, and it comes down to something I would not have expected: memory. Or, more specifically, is it worth remembering the creation?
One of the frustrations that I’ve had on 𝕏 is the limited amount of views my better photos receive compared to the images that are obviously AI-generated. Accounts that were created just weeks before for the sole purpose of auto-posting these fake images amass thousands of followers in short order. The total amount of effort put into the images comes down solely to the cost of compute, as an AI can generate a list of images to create, then create the images, while an automated system posts them to a given account on a schedule. No humans are required after the initial process is started. This used to upset me because it seemed unfair that a person with no involvement in the process could attract a large audience. However, once I started questioning the value of social spaces like 𝕏, the cost-benefit ratio became clear:
AI images cost “nothing” and there is no long-term value of the post.
The people making the automated systems do not care about the images, the posts, or even the followers. All they seek is the numbers required to receive a little payout from the platform. A similar pattern can be observed by those who like and follow accounts that make these posts, as they’re not thinking about any perceived value of the content. They see a nice picture, respond accordingly, and move on. The ephemerality of the moment barely registers.
Taking all of this into account, my frustrations with social platforms is focussed on the wrong problem. If I’m investing a tangible amount of time, money, and effort into something before sharing it in an ephemeral space, the problem is with my expectations. A photo can take minutes – or hours – to compose and capture, but only a split second to observe and scroll past. An article can take hours – or days – to write and rewrite before it’s published, but requires the reader invest minutes to consume. In an ephemeral space, minutes are an eternity. Most people will not invest as much into a creation as its creator … unless they’re married to them. Rather than be upset with “the algorithm”, I should be upset with my misunderstanding of the medium.
What does my misunderstanding of social platforms have to do with AI? It comes back to the question of whether the thing being created is worth remembering. On social platforms like 𝕏, Instagram, Facebook, and the like there is very little of significance shared. Most of it is momentary; important or interesting right now … but not tomorrow. If a creative person wants their creative efforts to be of value, then they need to use the appropriate medium.
Little jokes that can be cycled and repeated over time are perfect for 𝕏. Short videos featuring absurd situations or excessive drama are ideal for Instagram and TikTok. Articles replete with corporate buzzwords fit perfectly on LinkedIn. But this does raise a question: where does a person who wants their creations to be slowly savoured share their work?
Personal websites are a natural fit for people who want to maintain as much control over their creative efforts as possible, while reducing many of the distractions that come from using a crowded platform. There will still be challenges with discovery and reach, of course, as more search engines evolve to keep people on their sites and apps as much as possible, but personal sites that make proper use of the Schema.org metadata structures and llms.txt formats will see more visitors coming from the large AI providers, such as ChatGPT and Perplexity. It will be a slog, though.
The next few years will likely see creators invest in making their websites more discoverable to search companies and AIs. There may even be some experiments tried, with new social platforms being created to address some of the problems that come with rapid-scrolling, AI-generated content, and the like. A project I’m working on at the moment, while not social, will hopefully address some of the challenges that writers face in these interesting times.
The "10,000 hour rule" suggests it takes approximately 10,000 hours of practice to become a world-class expert in a field. While popularised by Malcolm Gladwell in his book Outliers, the idea is rooted in research on expert musicians. Some artistic pursuits may take more or less time to master.