Yesterday I wrote a post suggesting that new educational institutions will soon emerge, with AI at the centre of every aspect of the process. I recognise that this might feel like a step too far for some, so thought I’d expand on that claim here.
There are a few reasons for why I think AI will be at the centre of everything. And yes, I do mean everything.
AI is useful. The more we experiment with generative AI, across a wide-range of diverse tasks, the more useful it appears to be. I use it every day and every day I’m surprised at the utility it has for the work I do. This will exert a pressure to move us in the direction of using more AI for more tasks.
AI is getting better at everything. Plugins, APIs, and the new GPTs from OpenAI mean that we can connect foundation language models to 3rd party services, databases, and other AI-based systems. Previously, generative AI could talk about physics but connect it to a physics engine and now it can do physics. Connect it to Wolfram|Alpha and now it can do high-level computation.
AI is multimodal. When generative AI was predominantly text-based, it’s utility was limited to comparable text-based human tasks. The latest version of Bard enables you to interact with a YouTube video. ChatGPT can see, hear, and speak. Think of all the instruction-based content available on YouTube, now accessible through a conversational interface.
AI is simple. The fact that we can interact with the most sophisticated software ever created, using natural language, is a profound leap in terms of how we interact with machines. We are moving away from the keyboard, mouse, and screen as the user interface with computers. The Humane AI Pin is the latest example of an invisible interface, which I think will be come the dominant paradigm for interacting with machines.
AI is everywhere. OK, maybe this isn’t true yet but it soon will be. Generative AI is integrated into the next version of Windows. And all of Microsoft 365 (i.e. Teams, Word, Outlook, etc.). It’s being developed to run locally on your phone (i.e. no internet required). It’s being built into cars.
Now, take a moment to think about what you spend your time on at work, and whether any of those tasks are not exposed to AI. Even a simple conversation between two people in the corridor will have been mediated by AI at some point (because it will have had an influence on many / all of our previous activity in the day).
In this context, where AI is everywhere and embedded into everything, I start to feel like it doesn’t only help me to get better within my disciplinary context. AI helps me to get better at getting better. In other words, AI accelerates my learning. But this is more difficult when we try to force AI into existing processes and tasks, and where AI is positioned as merely another tool. When we tack AI onto business as usual, we miss the opportunities made possible if we redesigned systems in a world where generative AI exists.
Institutions that restructure tasks to become AI-first will outperform those who don’t because they will learn faster.