It has been a long time since my last post in 2023. It has just been really busy to say the least. I have been looking for a new physical home closer to the office where I work. We never had remote working during Covid. Then there has been travel for business and for pleasure. And the never ending yard work, which I hope to reduce or eliminate when moving to a new home.
In my last post I talked about gen AI and its impact on design. Here we are going on 2.5 years later and things have certainly changed. Many more voices are, if not saying design is dead, that it has been changed dramatically. I continue to dabble in it, but it has honestly not been a complete game changer like desktop publishing was in the 90s and the Web in the 00s. Yes, you can crank out designs at warp speed, but generally they are mediocre at best. They do not in any way really break new ground or solve problems on their own. Things are just faster… And as I noted as a probability in my previous post, the costs have been rising.
[continued from intro]
Salesforce just noted that their Anthropic bill was $300 million over the last year. As many other orgs are doing, they are laying off people estimated to be somewhere around 5,000 over the same period. They indicate these were mostly customer service roles and cited AI as the reason. However, the economics are beginning to look like they are break even at best as that would be around $60k per person replaced by AI. Which, depending on where their customer support folks were located, is more than the salary of an average customer support human.
Of course there are health and retirement benefits and other costs to be added to the human salary, yet the AI services are only beginning to increase their costs. To my knowledge not one AI provider is breaking even on their costs. At some point that will have to change, and so what the true cost for AI will be remains to be seen. Additionally, it is not clear if the AI is providing better, worse, or the same level of satisfaction as the humans were. There could be added costs in losing customers who are turned off by irrelevant, unclear, or outright hallucinated answers.
So back to AI and its impact on design. Most designers I see on the cutting edge are using AI to help them build fully functional prototypes, if not actual products. While this is great, what I am not seeing is revolutionary or even innovative design if it was not already the ideal from the designer. AI just builds what it is told. Although I am seeing some design folks trying to prompt the AI to be innovative. However, that is like asking a dog to be innovative.
Still, I do think this is huge as designers can generate prototypes do some evaluative testing with real humans (Using AI to act as human proxies is to put it kindly, idiotic) in quicker iterations than they have before. This certainly increases the possibility of discovering what would be an innovation that works. So I do recommend every person involved in UX to be exploring how to leverage AI in ways that help you build new concepts and evaluate them faster.