My top three ways for organisations to think about using Gen AI

Posted on 9th December 2024

Written by Andy Bell, CEO, FirstHelp

Finding the right language to talk about a new invention is an important step to understanding it. That’s why I love Drew Breunig’s essay “Gods, Interns and Cogs”:

Drew divides AI use cases into 3 buckets:

Gods are the human-replacement use cases. The much hyped artificial general intelligences (AGI) that are allegedly just around the corner
Interns are the copilots… Their defining quality is that they are used and supervised by experts. They have a high tolerance for errors because said expert is reviewing their output, and can prevent embarrassing mistakes from going further.
Cogs are comparable to functions. They’re designed to do one task, unsupervised, very well. Cogs have a low tolerance for errors because they run with little expert oversight.

I have been helping organisations think through their AI priorities and the ‘God’ vs ‘Intern’ distinction is very helpful. So I thought I would share my top three ways an organisation can think about using GenAI.

1. Avoid ‘God’ workflows

In thinking about how to use Gen AI I believe the first step is for everyone to suggest ideas. You will typically generate a dozen or so Post-it notes, of areas where AI could make an impact.

Since reading Drew’s essay, my next step is to divide the Post-its into ‘God’ ideas and ‘Intern’ ideas.

Without the Gods/Interns distinction, it is easy to slip into discussing God-like possibilities, where the AI is expected to act like an oracle. This pushes AI beyond what’s achievable, especially in casework environments where you are making important decisions based off pretty messy data.

2. What’s a good job for an intern? ‍

The second question for any organisation wondering where AI can deliver the most impact: If you had unlimited interns, what would they do?

That focuses the conversation on the job to be done and steers it away from pie in the sky.

What are the characteristics of good jobs for interns?

Time-consuming to carry out, but quick to evaluate
Ideally you can tell at a glance whether the intern has got it right. Drew says the ‘defining quality [of interns] is that they are used and supervised by experts‘. Ease of supervision is key.
‍
Safe fall back. Not critical if the intern makes a mistake
You wouldn’t trust an intern with life and death decisions.
‍
Lots of examples to copy
From the AI implementation perspective, this helps with training and with evaluation. It also means you are working on frequently recurring tasks, so it is worth the cost of automation.
‍
Drudge work
Some jobs are time-consuming and non-critical and frequently reproducible but they still aren’t suitable for AI powered workflows. In my own recent work with a social services organization, I could see that certain time-consuming tasks actually served an important purpose: they provided caseworkers with essential thinking time to formulate their recommendations. Perhaps not all “drudge work” should be eliminated.

3. Remember digital product development good practice

Finally, AI is a tool. Framing AI as an intern keeps product development discussions grounded.

Much of work implementing AI is in the realm of digital product development. LLMs fit into a standard toolbox: techniques like UX, MVPs and Agile all still apply. That’s not to say that nothing has changed. Large Language Models bring this unfamiliar, non-deterministic, automated intern to be integrated into software.

A big new challenge is going to be evaluation of its impact and ROI. This will be a big part of how we consider GenAI in 2025.

That human-ness of an intern, wired into software, is an extraordinary new capability for humankind. Over the next few years I believe it will become clear that AI is a huge help to caseworkers, saving time on admin and allowing them to better focus on the human connection which is at the heart of their job.

The future of AI in social services isn’t about replacing human judgment with artificial intelligence – it’s about giving professionals the digital interns they need to spend more time on what matters: helping people.