So I’ve been back at work just over a year now, and one of the things I’ve been working on is how AI (specifically large language models, LLMs) can best support me in my work. The two years I was off work due to illness was just as LLMs entered the mainstream, first ChatGPT and then many others.
Overall, its been harder than I expected to figure out where AI is actually useful!
I’ve gone down some rabbit holes, spent longer than I would like to admit arguing with Gemini (Google’s LLM, the main LLM I’ve been using as it's included on our Google Workspace subscription) and at times wondered what all the fuss is about.
I like to think I’ve now come to a bit of a happy place, understanding where LLMs can save me a bunch of time and effort, and where human-done is best.
Some of it I found really surprising, like that AI isn’t very good at adding! It was much faster and more accurate to create my own pivot or summary tables, rather than have AI do it and have to cross-check every calculation.
Other times, AI let me do things I would have really struggled to do on my own. For example, creating clean, bespoke visuals with three dimensions in Excel and rapidly building a survey in a semi-complicated platform I’d never used before, including use of coding. I still need some level of skill and understanding to do these tasks - having a concept for the chart I wanted, and knowing what survey features to ask for - but using an LLM saved a huge amount of time by cutting out most of the trial and error to create what I wanted.
Another thing AI has been really useful for is contract reviews. I still read draft contracts in full, but ask Gemini to identify any potentially problematic clauses and potential red flags to make sure I don’t miss anything major. While there is still a time and place for lawyers, as a micro business its simply not realistic to get a lawyer to review every contract, especially as some are for very small amounts of $$.
Something I didn’t expect was how much you can get from LLMs without actually giving them access to your data, for instance asking it to design specific formulas for a spreadsheet and then just pasting it in, without the LLM ever seeing the underlying data.
Of course, LLMs don't actually have a brain. I learnt this the hard way when an AI-suggested 'holding page' in my survey for people not yet ready to complete it unintentionally deactivated survey links—defeating the entire purpose of the holding page!
My other main useful uses of AI have been editing (including this blog) and brainstorming, particularly when I have an idea of what I want to do (e.g. intended outcome of a workshop activity) but am looking for a process or activity to get there.
I don’t think my job in monitoring, evaluation and learning (MEL) is going to be totally replaced by AI anytime soon. But I can certainly see how our role will shift even more into the facilitation and sense-making space (one of my favourite parts of my job TBH) with less time spent on document review and data cleaning, and maybe eventually data analysis.
What about you? I’d love to hear if your AI experiences gel with mine, and particularly if you’ve found other useful uses that I haven’t covered.