The Power of Instinct and Context in AI

Originally published on Research Live and Impact magazine in September 2023

Bethan Blakely, Research Director for Boxclever comments on the importance of instinct and context in AI

It seems wrong of me, as the analytics columnist for Impact, not to mention AI. I’ll be honest – I’ve been skirting around the topic for a while. The reality of writing means that my content is due to the editor about two months before it’s printed – and that can be a real sticking point when it comes to topics that move fast. AI, as we’ve all seen, is certainly one of those topics.

On the news the other day, ChatGPT and AI, in general, were described as the “new industrial revolution”. That’s big. Some things come and go in this industry, you hear buzzwords being thrown around, and then suddenly disappear. AI is not one of those buzzwords. In fact, the noise just seems to be getting stronger and stronger, with more people asking questions and wondering what the fuss is about (and a large proportion wondering if they’ll lose their jobs to it anytime soon).

Let’s start there. As far as I see it, no, you will not be losing your job. The number of people employed in the research and analytics industries may fall by a small percentage. But you will not walk into the office (or join a zoom call) and see a robot sitting where Pam used to sit. Generative AI tools (that is, AI tools that can generate new data themselves) will become normalised in the industry for their ability to streamline certain arduous processes – making you more efficient at your job. (This is where the small percentage decrease may well come from – we might end up with a team of 28 instead of a team of 30 because everyone is more efficient.)

Having said this, there are multiple reasons why a generative AI bot will not be sitting at your desk doing your job for you anytime soon. Turns out, being human has its advantages. I recently read an article that explained, in scientific terms, why a 2.5-year-old toddler is smarter than generative AI. AI uses a large dataset to detect patterns, and then generative AI uses those patterns to generate new data, which is similar to the old data it’s already investigated. Toddlers (and humans in general) don’t need a large dataset. You can point to a lake and say “look at those boats”, and, because of the human ability to understand context, a toddler would clock what a boat is. An AI tool would need to be fed thousands of data points explaining what a boat is before that link is made.

As a human, you’re also much better at interpreting analysis and results than an AI tool. You can look at a headline and your gut will tell you “nah, that can’t be right”. This instinctual feeling that comes from your expertise combined with your personal experience and the context of the situation puts you in a much better position to understand whether things “feel right” or “make sense”.

Another way in which you have the upper hand over AI is in disseminating and communicating the findings of your work. Yes, you may well use generative AI to help structure your story, or proofread your work, but you know your audience better than the AI. You know the tone of voice that’s appropriate, how to land a difficult message, and the context that’s needed.

All in all, your odds are looking pretty favourable when stacked up against all of these snazzy new tools. That doesn’t mean it’s time to sit back and relax.  In fact, very much the opposite. Glance at an agenda for any high-profile event in the research and analytics space now, and you’ll see several talks on AI, ChatGPT, the metaverse, and more. People are getting to grips with these tools and making them part of their day-to-day. This means that if you’re not – you will eventually get left behind. The good news is that there are countless resources online, many of which are free, to learn the ins and outs of these tools. So get out there, get your inquisitive hat on, get your hands dirty. Download these tools and have a play for yourself. Get involved in conversations and hear how others have been using them and learn from their mistakes.

You might be better than these tools in isolation – but you’ll be even better again once they’re a part of your everyday toolbox.