At a glance

AI multiplies what you already have. If you have judgement, it makes you considerably faster. If you do not, it makes you faster at being wrong.

Multiplication is the right word, and it is worth taking literally. Multiplying by a large number is excellent. Multiplying by nothing gets you nothing, at speed.

What it actually accelerates

The typing. The recall. The tedious middle part between deciding what to build and having built it.

For a developer who knows what the answer should look like, that is a serious gain. You skip the boilerplate, you skip looking up syntax you half-remember, you skip the forty-minute detour into someone else's documentation. What used to be an afternoon is an hour, and the hour is spent on the interesting part.

We would not give that up, and we do not think anyone should.

What it does not touch

Deciding what to build. Deciding how it should be structured. Noticing that the requirement, as stated, contradicts itself. Knowing that this shortcut is fine and that one is a trap.

That work did not get faster. It cannot, because it is not typing. It is the part where someone holds the whole problem in their head and makes a call that they will still stand behind in two years.

AI produces an answer to every question at the same speed and with the same confidence. Deciding whether the answer is any good takes exactly as long as it always did, and requires exactly what it always required.

Why this cuts both ways

Here is the uncomfortable bit.

The developer who understands the system reviews the suggestion, spots that it has quietly ignored the edge case, fixes it, and moves on in seconds. They got faster.

The person who does not understand the system accepts the same suggestion, because it looks reasonable and it runs. They also got faster. They are now generating decisions they cannot evaluate, at a rate they could never have managed by hand.

Same tool. Same output. Opposite outcome, and the difference is entirely in the reviewer.

The line worth remembering

A calculator makes an accountant faster. It does not make you an accountant.

Nobody finds that controversial. For some reason the software version of the same sentence still causes arguments.

So what should you take from that?

If you can review the output, use everything. Push it hard. It is a real advantage and the people refusing to touch it on principle are simply choosing to be slower.

If you cannot review the output, that is fine too, as long as you are honest about what you have: a fast route to something that works, with nobody in the loop who could tell you whether it is sound. For a prototype that is a perfectly reasonable trade. For software people depend on, you need someone in that gap.

We have written the fuller version of this decision here: should you build your software with AI?

Or in the words this series started with: AI is a brilliant typist and a terrible architect. Typing faster has never been the hard part of building software.

Not sure whether anyone reviewed yours? Our AI Code Audit is a fixed-price £495 review that tells you plainly what you are dealing with, or get in touch for a no-pressure conversation.