Make AI useful with a personal audit

Author

Calvin Brown
April 18, 2024

Welcome to AI Communicator!

Today’s issue is the third in a three-part series on getting started with AI for communications:

Make AI useful with a personal audit

You’ve picked your Large Language Model (LLM) and more or less mastered it. 

But what now?

One of the frustrating things about AI - and there are a few - is knowing that you have your hands on something incredibly powerful with amazing potential, but not knowing quite how to unlock it.

It’s definitely fun, impressive, and sometimes uncannily brilliant.

But it can also feel like a solution in search of a problem.

I have this amazing tool at my disposal. I should be able to accomplish amazing things now - with next to no effort. 

Or at least move beyond using AI as a toy and towards it becoming a useful digital assistant.

This issue of the newsletter is an attempt to start to sketch how to do that in a way that is tailored to you. 

The focus will be on individual AI use. Team, department, and organisation-wide adoption are different topics that will no doubt crop up in future newsletters.

So how best to fit AI into current workflows?

A really important thing to consider upfront is that current LLMs need to be micromanaged. 

They can be highly capable at individual tasks, but don’t expect them to undertake long and complex projects without instruction and guidance at each step. 

Effective autonomous agents with these kinds of abilities are expected to be one of the next big things in AI.

But right now, your own judgement and professional expertise is important to carefully managing the AI to the desired result.

That is why delegating to AI works best when you have an existing documented process, standard operating procedure or style guide to prompt the LLM.

Bearing this in mind, start by looking at how you have already been using AI.

What other tasks do you do regularly that AI could help with?

Consider your current and upcoming projects and any potential uses of AI.

Consider your task inventory and look for pain points. Where are the inefficiencies, repetitive tasks or bottlenecks?

Some of these might be small things, but if done frequently the benefits of AI add up over the days and weeks.

Think about things you personally don’t enjoy, or can’t do well.

Maybe you love brainstorming, but struggle to choose the best idea. This could be an opportunity to use AI to help you critically evaluate your ideas.

Or perhaps you struggle to come up with any ideas, but enjoy considering other people’s.

Let the LLM generate ideas, something they are good at, and you can choose the best.

All this should build  a good starter list of AI uses tailored to you.

You can, of course, give your list to your LLM and ask it for more suggestions.

As you start putting these ideas into practice, keep track of how much time you think it is saving you and how the quality of results compare to doing it yourself.

Not that this is about immediate benefits.

For anyone starting to implement AI in their work, learning is still a large part of the process.

And we shouldn’t rush to dismiss things that AI doesn’t do perfectly, or as well as we’d like, right now. 

This is the dumbest LLMs are ever going to be. 

Sam Altman, CEO of OpenAI, said recently of GPT-4: “I think it kind of sucks, relative to where we need to get to and where I believe we will get to. I expect that the delta between [GPT-]5 and 4 will be the same that was between 4 and 3 and I think it is our job to live a few years in the future and remember that the tools we have now are going to kind of suck, looking backward at them, and that’s how we make sure the future is better.”

Time spent on LLMs now will be a very useful investment. 

It is likely that the use cases and solutions you are learning and developing will produce outstanding results in the future instead of merely good.

Until then, embrace the suck.

I’d love to hear what you think about the newsletter - or anything you’d like to see featured in future issues. Reply to this email with feedback or questions.

Thanks for reading.

P.S. I have a favour to ask. If you are enjoying the newsletter, please share it with colleagues who might be interested. They can sign up for the newsletter here.