So What?
But with AI now!
I first heard this phrase at Amazon while working on Alexa. I was presenting a proposal for a revamp of the BI systems my team owned when one of the principal engineers in the room asked me that question. When I heard that question, I was completely taken aback. I have never heard a question like that at any workplace prior. You propose something complicated only to be met with a vague, somewhat rude question?
So when I heard that question for the first time, I asked him to clarify what he meant. He replied, ‘Ok, so you execute and deliver this project. What happens next? Why does this work matter? How does this help customers? So what if you do this project? And if you don’t do this project, what happens? How does this affect the business?’ Then, just like that, it all clicked in. I had focused solely on the details of the project, how to design it, staff it, and execute it, but not so much on the impact the project could have on the end customers. The ‘So What?’ question was designed to get me to think critically about the actual implications of doing (or not) the project.
I have used this technique throughout my leadership and executive career. Asking myself the ‘So What?’ to all major investment ideas, thinking critically, and digging into the details of what will be the actual impact on the bottom line has helped me separate the chaff from the wheat. I also encourage my teams to think critically about their proposals, including clearly describing and quantifying the impact before submitting them to me, because they know I will ask the ‘So What?’ question, although I may not use those exact words.
Interestingly, I have found myself using this technique more frequently to cut through the hype and frenzy surrounding AI. In this post, I am going to take a few hype statements and try to cut through the chaff by using the ‘So What?’ technique.
Hype
Google’s Veo model can now create realistic movies from text prompts!!
Reality
Will it replace human filmmakers? Nope. Models are trained from data created by humans. Data pulled out of art, movies, and films made by humans. But here is the thing, though. AI models always narrow down to the commonalities in the dataset they are trained on. They are great at picking the most commonly accepted answer to a question. Essentially, it propagates the ‘sameness’ in the dataset. Unless you want to watch AI-generated movies that have extremely predictable visuals and stories, this will never replace real-world filmmakers. Will AI make it easier for filmmakers to make movies? Absolutely. It will take away the grunt work. However, creativity will need to be provided by humans.
Hype
Co-Pilot/Claude is getting crazy good at writing code from text prompts
Reality
Most software systems used by companies are deterministic in nature. AI systems (including GPT-based ones) inherently are not. Simply put, a calculator app should always output 4 when asked to calculate 2+2. Someone has to be able to course correct the AI-based coding system if it starts hallucinating, which it will. There will always be a need for a trained human to oversee what the AI system is doing.
Another fact is that sometimes the requirements are just wrong. The product manager might be thinking about the feature incorrectly, or could have misinterpreted the customer feedback, or the customer feedback itself may be incorrect. It will take a human to figure out if the thing the team is building is actually the right thing. Again, someone has to be at the wheel.
Lastly, the notion that AI will eliminate the need for mid-level engineers is simply incorrect. Senior engineers are promoted from mid-level engineers, who in turn are promoted from entry-level engineers. Maybe in the distant future, the overall size of the organizations will become smaller, but software engineering as a discipline isn’t going anywhere.
Hype
AI will replace customer support reps
Reality
Firstly, AI will not replace customer support representatives. Klarna replaced about 700 customer support reps with AI-based chat/voice systems, and it failed. Klarna’s CEO admitted that the automated systems were providing a low-quality service to customers. Now they are looking to hire real people back.
Getting answers to your questions is only one of the reasons people reach out to customer support. The major reason people continue to press 0 or yell, ‘I want to talk to an agent,’ to automated systems since the early 2000s is that humans understand other humans. A human can understand nuance and undertone. A human will break the rules to make their customer happy. They can lend a sympathetic ear. They can provide workarounds. Will AI reduce the overall size of CS teams? Absolutely. Will it replace humans? Not until AI develops consciousness.
Hype
AI tools will remove the need for branding and creative teams
Reality
Think about the most iconic brands in history. They stand out. AI tools that are built on top of existing data can never stand out. They propagate the sameness. If you let an AI tool design your brand, colors, and copy, it will be more of what already exists. Your brand will never stand out. The creativity that makes brands pop and copy come alive needs to come from a human. You think ChatGPT could have come up with the Nike swoosh or Apple’s Think Different tag line? I don’t think so.
Hype
ChatGPT can write sales emails as well as humans
Reality
AI-generated sales emails will end up in the same place you send all the human-generated ones. The current conversion rates on outbound emails are ~2%, and AI-generated sales emails will probably have a lower conversion rate.
Every time I apply my ‘So What?’ technique to all these hype moments, it becomes clear that most of the noise out there is just hype. And the people generating the hype are the ones who will most profit from it. A.K.A., the investors, founders, and companies who are heavily invested in this hype.
To be clear, the current crop of AI tools is not to be scoffed at. Their capabilities are phenomenal. They are great at filling in the blanks. They excel at automating manual labor or repetitive tasks. They will most certainly reduce the size of certain teams pretty significantly. However, they won’t build the next iconic brand. They won’t build the next iconic product. They won’t write the next bestseller. They won’t invent the next wave of software development technologies. To do those exceptional things, you need creativity, randomness, and a touch of chaos that can only be provided by humans. So next time you see a hype article on social media, ask yourself, ‘So What?’