It’s funny how we have tech and automate a lot of stuff, add some monitoring for the tech, then employ people to watch those monitoring tools. It’s like the more we have or develop, the more we lose sleep over, especially companies that want to stay ahead of the curve or protect massive data. With the introduction of AI, it’s not even getting better, especially for industries which parade and wear it like a badge: “We use AI to blah blah blah…” Firstly, I think only a small number of the AI market benefits the end user, and companies who depend on real humans for revenue should not be singing AI at the top of their lungs. After all, they’re serving humans, not robots.
A thorn in one Bolt driver’s side in South Africa caught my attention. The driver was frustrated by the sudden closure of Bolt’s physical offices and moving their services towards AI. Folks, we know what that means: removing humans to leave customers to interact with AI chatbots. The poor driver was left with Fred (AI chatbot), frustrated with the lack of context comprehension to his service request and not pleased with the tone detection instilled in Fred, which gave readings that he’s angry when interacting with Fred. As a bystander, that is not a good move. Starting from the top, being a driver is not easy—from customers to other drivers on the road causing chaos throughout your day. Then AI is not good at detecting different English dialects that are influenced by African languages. We do not speak the Queen’s English in Africa, as our languages are rich in expression and it’s hard being constrained to the English language pack. Back to Fred: I felt bad for the poor driver as I’ve had my fair share with AI bots calling me and trying to use them to search for information (score still zero, never won). Bolt should ask Duolingo what happened when human-serving employees were removed/reduced from the equation—definitely the other side didn’t balance at all.
It is 00:13. I am writing this because of a two-hour session I had speaking to Google Gemini. Please don’t judge me; I won’t judge you. What I have concluded reminds me of a few years back when I was in the integration space. Our project management tool (not dropping names, but it starts with J, ha ha) was very slow. For several weeks there was an investigation. In the end, one team created a script that went berserk and created millions of insertions to the tool, which is in a shared space within the different units. This was hilarious for me, because all these tools we create keep us up at night, and automation—when it goes wrong, you won’t know immediately, as it’s watching the system, but there is no one watching the watcher. And if there was, then who will watch the watcher of a watcher? See where this is going? Hint: Tower of Hanoi (recursion). Guys, we’re going to lose sleep over this. Can we just think for a moment: who are we serving? And insert AI where it makes sense, with robust guardrails that do not impose themselves on our finances.
The cloud, AI, and rushing to modernize everything is just hurting real humans. We need to take stock of ourselves and think thoroughly: is this for good (including morally good profit) or experimental with no blast radius contingency? After all, we all serve humans and we are also humans.
