4 Comments

Very interesting article thanks. At the start you pose the question "So if you’re a larger organisation that has some capacity to start exploring, how should you decide where to invest?" -- are there characteristics in common across the projects you mention (and other relevant examples) that could help answer this question with some general principles? What makes a "safe bet"?

Expand full comment

Good question. I would say for the most safety you're looking for a few properties, something like:

1/ the process is done frequently

2/ the output is supervised

3/ the potential downsides and also upsides aren't too large

4/ you can easily compare against the vanilla process

5/ technical outlay is small

6/ other organisations have seen promising outputs with similar things

7/ you're sure there's no "off the shelf" tool that will do what you need already

Most of these are things that you'd want for any kind of pilot, not just AI!

Also, certainly there will be riskier projects that might provide more benefit, but it seems like a good idea to start with the basics, and also - given the speed of change - wait for the private sector to make all the mistakes first, and also develop the tooling.

Expand full comment

Thanks, that all sounds very reasonable. Perhaps something about being primarily internal facing tools too?

Expand full comment

That's what I'm calling "supervised" essentially.

Expand full comment