Fibres, wet pads and plastic

Is embarrassment AI's biggest barrier?

"But you didn’t order fibres."

I was standing next to a concrete truck driver going through the spec of my second concrete delivery. This time to make a concrete slab: the floor of my workshop. Modern concrete trucks are amazing contraptions. They mix the concrete on site, removing the need to try to accurately estimate concrete volumes. But if they don’t have the right ingredients...

Fibres are little bits of plastic added to strengthen concrete. Concrete shrinks as it cures and o3 had decided fibres were a cheap way to reduce surface cracking. So I took o3’s advice and got them added. But, somehow, they hadn’t made it onto the driver’s order form.

"You’ve got rebar so you don’t need fibres," chipped in the pump operator. This was someone with decades of experience of pumping concrete. Decades of experience of laying slabs. What he said seemed to contradict o3. I paused. Should I mention my expert was o3? That o3 had access to the best engineering practice? To the best structural engineers? Would he even know what o3 was? And if so would he dismiss it out of hand?

I decided it was best to avoid mentioning o3, and muttered something about having read it somewhere. Technically true, but definitely not the whole truth. At that moment the concrete truck driver re-appeared holding a big bag of fluffy white bits of plastic. It turned out he had some fibres left over from a previous job, so we were all set.

Later as we started to pour the slab, the pump operator noticed the wet pads I’d setup earlier that morning. Wet pads are little mounds of stiff mortar set up around the pour site. They are set to the finished height of the slab, so make it easier to get a consistent level as the concrete splurges out of the pipe. "Good idea, those," he nodded approvingly. Then added, "Never seen them before".

Once again I had o3 to thank for suggesting them.

Reflecting on that morning, I realised I’d learnt two interesting things.

The universal domain expert

First, AI is a remarkable domain expert. It clearly knows more than the professionals. The people who’ve spent their careers working in a given field. It knows all the tricks and tips. That shouldn’t come as a surprise. o3 has been trained on massive volumes of structural engineering. It has consumed more information on concrete than I could ever hope (or want) to. Learnt the theory. Read about countless real world examples.

As I write my wife is researching kitchen appliances. She’s wrangling o3, 4o and Claude. Earlier this week she got o3 to do a detailed analysis of a quote for a new kitchen. To help her understand what’s good value, and what’s not. She’s moving beyond the point of AI just enabling her to do something faster and into the territory of doing new things. In the old world doing this sort of analysis would have been hard and time consuming. Now? It’s just a few prompts.

At this point the humans who do not involve AI in their daily life are starting to fall behind. They are putting themselves at a disadvantage. The reality is AI can help with pretty much everything. You have nothing to lose by asking for its opinion. Listening to its suggestions. Learning from it.

Embarrassment

Second, it can be embarrassing to talk about AI. It’s a controversial subject. An emotive one.

I meet sceptics on a regular basis. The people who tell me Claude is useless because it can’t reliably tell whether the number 9.11 or 9.9 is bigger. Sure - AI can struggle with things we humans find easy. Sure - the frontier is jagged. Sometimes AI is amazing, other times it face plants hard. But that doesn’t mean we shouldn’t use it. Just as I wouldn’t use my cement mixer to make a cake, there are things AI isn’t good for. Figuring those out is all part of the fun.

But admitting I use AI to another human? That can be tricky. Admitting my concrete knowledge came from o3 was unlikely to gain me anything. But I stood to lose. There’s a time and place for evangelism - and that morning wasn’t it.

One of my colleagues recently told me they were embarrassed to tell others they’d used AI on a task. They felt that using AI somehow de-valued their work. That they had cheated. When the reality was they’d freed up their time to do the things only they could. AI shame - people feeling embarrassed to admit they used AI, and people who ridicule others for using AI - is real.

At the time, as an outside observer, I knew my colleague should be proud of using AI to save time, not feeling embarrassed. I was proud of them. Yet now here I was, standing in my wellie boots with concrete gently gurgling over my toes, embarrassed to admit my 'expert' was an AI. Even though o3 evidently knew more about concrete than the professionals around me.

The real barrier

The irony isn't lost on me. The tension between AI's capabilities and our social discomfort with using it may be the biggest barrier to adoption. Not technical limitations or data privacy concerns, but simple human embarrassment.

And my slab? It’s gently curing under plastic sheeting. Something most professionals don’t bother with, either due to ignorance or laziness. But another suggestion from o3 designed to ensure the concrete cures properly. Ask me again in a week if it was worth it!