Disclaimer: AI has not been used to create any part of this blog - this is entirely conceptualised and written by Mac Alonge

That feels like a super weird caveat to start with, but the sad reality of the situation is that AI is being used in almost every facet of our lives. Where our immediate thought, in most situations, would have often been to think things through, ask a friend or use a search engine, most of us now turn to our favoured AI assistant.

It's weird how much the world has changed in a seemingly very short time frame. Weirder still, is how much it is continuing to change. This feels significantly different to historic technological advancements, as the logical conclusion seems to be our complete redundancy. I genuinely fear for the next few generations, including my kids - What jobs will they have? What will work look like? So many questions and so many unknowns. But that's one for another day. For the time being, if AI is here to stay (which it looks like it is), we must navigate how, when and why we use it. 

Over the past few years, we've seen AI being increasingly used in relation to EDI, and whilst I understand the benefits and rationale, it's been hard for me to bite my tongue. My problem is that I've seen AI often used clumsily for EDI, where the user doesn't fully appreciate the potential harm that they are unleashing. 

A lot of focus in calling out the potential harms that come from an AI centric approach centres on the potential to hard code algorithmic bias and further entrench disparities. But "the computer did it" is only one part of the minefield, there is so much nuance to the pros and cons of using AI. . 

Below are a few examples of where I've seen people struggle with using AI effectively, not necessarily distinct to EDI, but food for thought all the same. 

1) We recently lost out on some word to help an organisation that we have established a reasonably good relationship with over the years. Their brief was very ambitious. We proposed to deliver a piece of work that would have been more impactful, but didn’t meet their expectations in terms of scale. They ended up going with an organisation that had claimed to be able to deliver the full scope of their request with less than the available budget. I caught up with the client lead a while ago and asked about the project. They laughed! They said the project had failed miserably and they were currently trying to salvage something from the project. In the world of AI, anyone can sound like an expert and deliver a polished pitch, but the reality of delivery, especially in the world of EDI is far from straightforward. 

2) Next, I was on a call with an enthusiastic young consultant, who had reached out for some advice. As we were talking I noticed that they kept glancing away and appeared to be reading questions to ask me. Of all the uses of AI in relation to EDI, I have very little issues with this, as it can be a great mechanism to learn. However, when deep thinking and curiosity are substituted with AI generated, generic fillers, opportunities to really learn, explore and lean into your intuition are lost. This is especially true for people earlier in their career, where honing your craft is key. 

3) Over the last few weeks we've been asked to review and consider pitching for projects that were clearly scoped or shaped by AI. The problem here is that, whilst the AI captured the sentiment and structured the request in a seemingly sensible way, it came across as rigid and superficial. When we probed into the rationale behind certain parts of the proposal, it was clear that the person's understanding of what was proposed was shallow. Moving from (AI) written request to in-person conversation demonstrated the gaps in articulating the initial problem statement and how the client had moved from that problem to the proposed solution. There were fundamental gaps in the solution development that the client couldn’t explain. This had the impact of exposing their lack of understanding of the subject matter, and the potential implications of the choices that AI had suggested.

If we end up in a world where AI gives us the questions and the answers, we really need to critique what our role in this equation is. If we outsource our thinking to AI, do we resign ourselves to inefficient administrators or matrix-style robot feed? 

Whilst AI is a powerful tool for greater efficiency, there are some areas where AI can fall short in the hands of the uninitiated.

Context - AI can be a great help, but very rarely do we have the ability to fully articulate the wider context within which we work. I think we often underestimate the role that our guts play in decision making and figuring out tricky issues. EDI is a wicked problem, for which the treatments are very rarely simplistic. If it were simple, everyone would be doing it well and society would be thriving and inclusive. Where AI works best is applying rules and principles that have been proven to work well over time. The issue with EDI is, in my opinion, there are very few interventions that work well for all organisations, at all stages of maturity, across all sectors and all company sizes. EDI is an evolving picture and AI doesn't have the same evidence base to determine a) all of the available options, b) the short and long term consequences and c) the contextual factors that determine the suitability of one intervention over the other. 

Process - EDI work is more about the process than the content. There is a lot that can be learnt by going through the process and iterating, then getting to the "answer" as quickly as possible. Humans are complex, so naturally it takes a while to fully play out all the implications and consequential considerations. Unfortunately AI isn't quite there yet, especially in the hands of people that just want to get to the answer as quickly as possible. EDI is not a problem that can be "solved" in a traditional sense, but it is something that can be managed in a careful and considered way. That takes skill, technical abilities and a touch of artistic flair that AI isn't yet used to. 

Infrastructure - "doing EDI well" requires mature and inclusive infrastructure. For too long leaders and organisations have assumed they can take a bolt on / tick box approach to EDI. Change the text to be more inclusive, add a couple of pictures of "diverse folk", apply for or sponsor a few awards - job done. But truly moving the dial forward requires the whole organisation's infrastructure (people, processes, systems, skills, strategy, structure etc), to be working towards delivering inclusive outcomes. You can't simply "brand" your way into equitable outcomes, in fact, such an approach would likely work against you, as staff pick up on the contradictory nature of what you say Vs what you do. 

My view is that we need to use AI to supplement the areas that AI cannot easily replicate - curiosity, interest, exploration and artistry. When it comes to AI, specifically in relation to EDI, we need to hone in on these areas.

The tendency now is for a lot of organisations to look to AI first, which is a trend I don't see reversing any time soon. However, for a lot of organisations this creates a dangerous dynamic where the foundational EDI knowledge has not yet been adequately built to ensure there is an understanding of what direction the organisation should be moving in. This means that any guidance is likely going to be generic. AI is a great tool if you can be precise about what you want and have an idea of what good looks like. When it comes to EDI, unfortunately most leaders and organisations can't answer either of those questions. 

My advice for those keen to adopt AI in relation to EDI would be to take some time to think about what areas of the business are mature enough to leverage AI well, and start there. This can be done while taking some time to think through how best to build the maturity of other areas of the business. 

One of the best things about AI is the ability to iterate and work through tricky issues. EDI is a wicked problem and the ability to iterate and work through potential outcomes and implications before implementing, should help just organisations to more accurately make meaningful progress. 

Lastly, AI should not be adopted without conducting a robust Equality Impact Assessment, to ensure that your organisation has full sight over the potential risks associated with AI adoption and a plan to mitigate any potential adverse implications. If you would like to learn more about how you can conduct a full equality impact assessment on your upcoming or existing AI adoption plans find out more here.

For anyone interested in exploring more about AI in EDI, join our conversation here.

Schedule Your FREE Consultation Today

Sub-disclaimer: whilst I didn't explicitly use a generative AI model to ideate, draft or review this article, autocorrect on both my notes app and Google docs was used ...the robots really are taking over!