This is a topic that has been on my mind for a while, and it felt like a good time to articulate what I’m feeling. Look, I love AI and use it on a day to day basis, but I also absolutely detest being handed a 40-page slop-doc or a prototype that looks like it was generated by GPT-3. I’m sorry, but if I see a purple gradient I will immediately tune out. There isn’t anything wrong with using AI to communicate your ideas, especially at lower fidelity, but I think it weakens your org’s brand if you’re presenting straight BS to customers and asking them to ignore the lack of detail entirely.
Now, I am liable to being naive, and I want to be cognizant of that, but surely there has to be a better way? I have long believed that if you are building anything remotely customer facing — a product, a website, developer documentation, even a prototype — you have to do the due diligence of actually thinking through the problem deeply and ensuring the idea passes the sniff test of being worth presenting. AI lets you bypass that thinking stage entirely and jump straight to something that functions well enough to get in a customer’s hands. The signal you gain from those conversations can be helpful, but it often misses the point entirely. I’ve seen too many AI slop prototypes get validated by customers when really we should be asking ourselves whether the idea is even worth exploring.
The cost of execution is starting to become nominal, but the cost of customer relationships fractured by this isn’t. Personally, I always try to hold myself to my own standard of polish before showing something to a customer. I understand that can differ at different stages of discovery, but I think the extra effort it takes to bring something up to brand standards is worth it.