Make AI prompt you...

Why the real intelligence breakthrough will come from flipping the AI relationship on its head

I had a conversation with a writer friend of mine the other day who, like most knowledge workers and creatives, is thinking about the ways AI will change (or even threaten) their craft and livelihood. Whilst they seemed generally conscious of industry trends, from a writers point of view they were pretty unfazed. “AI lacks that certain Je ne sais quoi” they told me. Another writer framed it to me as AI lacking a ‘soul’.

I think it's generally true that if you want AI to generate content for you based on your ideas, you're in for a disappointment. At first, the generated text seems impressive, but once you start to really read into the text you find it’s bland, algorithmic, and generally ‘safe’ to the level of being utterly dull. Anyone with even a little exposure to the tech can probably spot it a mile away.

However, even with this in mind, I feel a pull from the markets and companies towards exploring how this tech could save them money and up their throughput when it comes to writing. Right now it’s commercial content creators and ‘10 best washing machine’ blogs, but maybe one day soon it’s breaking news updates, then thought pieces. This technology is still in its infancy and despite sounding generic now, we’re probably on a path to brilliant creation and communication coming from AI.

For me this raised a question even more important than if my mate keeps their job (although that is of course important…). This transition is a part of a much bigger trend: if we keep edging towards a world where we offload this sort of cognitive, creative, generative work to these AIs then we run a real risk to human agency and thought.

Knowledge is refined, communicated, and shared by the written, the visual, and the auditory. We now have technology that can generate all of these things, but when does this impressive technological feat tip into disempowering creation?

Right now it feels this evolution is framed as only two choices: humans making everything (good) or AI making everything (bad). However, there is a third way - a light at the end of the tunnel that could reinvigorate this world changing promise of truly augmented human intelligence - and it starts with flipping one of the most foundational elements of the generative AI on it's head: instead of us prompting AI, we need to start thinking about AI prompting us.

Hit reverse

The word 'prompting' has now become all but synonymous with the image of typing instructions into a small AI chatbox. We type, and AI makes. This one directional view of instruction and creation is the current way most of us have been living in this generative AI world. However, I think we need to open this up and start thinking about both directions of the information stream.

I want AI to ask ME questions; I want AI to push ME to go further; and I want AI to get ME to 'regenerate'. WITHOUT doing the work for me.

This may sound strange to some, but when it comes to human thought, clarity, and creativity, this is usual practice: a therapist will prompt you to explore your own feelings, a coach would prompt you to dive deeper on your own problem solving, and a creative writing class will give you prompts as a starter for getting the juices flowing. So why shouldn't we think about using AI in this way?

The other day I was stuck writing a blog and I made a visit to Claude. I outlined what I was exploring and some of the opinions I had, but shared that in general I was struggling to distill the exact elements I wanted to focus on and the clear argument I wanted to make in this piece. I told Claude to ask me questions to help me explore this and explicitly made it clear that it should NOT generate any text itself beyond these questions. There was a little friction, but 15 minutes later I had 10x more clarity.

This isn’t some productivity hack or cheat sheet for AI in your workflow; it’s a fundamental question about how we use this intelligent technology to either remove cognition and agency or genuinely enhance humans abilities - part of a larger trend I’ve explored, viewing AI as a sense-making technology.

There’s a real opportunity here where AI could help far more people access the thought partner we could do with. Not only will this support the production of more critical, thorough, and through provoking content (and possibly out-race the slop), but it could play a role in a vital shift around our relationship with this technology - empowering people rather than creating a fear of replacement.

What the big players have wrong

Products matter. The products we build now will largely shape the way we think about generative AI technology and define humans’ relationship with it, yet many are falling pray to quick integrations and cheap feeling deployments.

Many current generative AI roll-outs for writing feel dumb at best, and dangerous at worst. Big players in the writing space (along with the many SEO-optimising trash generators now flooding the market) all have this main thing wrong: the go-to function should never be to immediately have AI generate text for you. It should, instead, open the dialogue needed to help authors find their own message and words. Can it generate some inspiration from time to time? Sure. But the immediate action should never be creation, it should be the bouncing board, the brainstorm partner, the devil’s advocate.

How we can do it right

If we want to re-think the role of AI in thought and creation, we have a well of good practice to build on - from coaching training, to therapist handbooks, all the way to the insights of a great book editor. Ask open questions, orient around action, help think about it from someone else’s perspective… there’s not shortage of tricks of the trade.

Right now, this probably won't impact the visionary writers of our time, but it could certainly help some of us mere mortals who actively want to ideate and communicate in the best way we can. As AI gets better, this progression will get better and maybe one day we really do reach the point where human intelligence here is augmented to go beyond its current upper bounds.

If I can summarise this is one sentence: we shouldn't be telling people AI is better than them; we need to show how AI can help them be a better version of themselves.

Reach out

This is something I'm actively exploring at the moment and I'm excited to share more soon. If you'd like to share your thoughts on what this future needs to look like, reach out and let's chat.