With the advent of AI, businesses and individuals have the capacity, to a greater extent than ever before, to be dishonest about their capabilities.
It is often repeated that AI enables a single person to do, in a day, what a company may take a week to achieve. However, this is only possible with a sacrifice in precision.
Whereas once, you would carefully consider every line of code, now you may not read a line of code. When done correctly, the overall structure is so well understood and described beforehand that one may be able to perfectly guess any given line of code (which is generally how LLMs work). However, this is often not the case.
AI does its best work creating convincing sentences, and Humans do their best work finding meaning. We must take an intentional stance against finding meaning in the meaningless.
A company is made of the people who work for it, not the product it sells. If the people who work for it do not build the product, it is not the company’s product.
Once upon a time there was a software engineer. This engineer hated their general manager because they felt that they spent more time in meetings than working on their code. The end.
The moral of the story is that two-thirds of our work is in knowing what to build. AI can, and should, only solve the one-third.
So we’d like to rephrase the saying: AI allows a single person, in a single day, to turn functional requirements into a product, whereas a company may have taken a week.
Holstegge Technologies makes use of generative AI. We would fall behind without it. However, we are sceptical of its true value beyond data aggregation.
There is no sin greater than publishing generated text, for then your own message is diluted.