Realising the potential of AI; Welcome to Partnership, Co-Intelligence and Micromanagement agentic-style 

A study in Ethan Mollick’s book ‘Co-Intelligence – Living and Working with AI’, characterises two ways to approach using AI; being a Centaur, or being a Cyborg. 

In this study, Ethan helpfully explores what the best ways to work with AI are. Spoiler alert #1, it’s not about telling people to ‘just get used to it’. In fact, it confirms what I think all of us in the HR community have instinctively known for a while, that AI is not an instant game changer. 

Truth is, without considering the cultural transition needed, and ensuring mindsets are aligned when working with AI, the potential could be harmful not helpful. But AI is here to stay and if it’s used in the right ways, it genuinely does have the potential to supercharge the output we are capable of. 

So, what are the ‘right ways’ to use AI? Spoiler alert #2, we want to be Centaurs, not Cyborgs and work co-intelligently with AI. 

In plain language this means keeping things clear-cut when it comes to who does what – humans and AI each have their own lane. We must be pretty strategic about it too, playing to each other’s strengths. Ideally, we’re working in ways that let AI support and enhance what we are great at, without stepping on our toes. It’s never about a total merge, it’s about being co-intelligent. 

Before I get to real world experiences I’ve had when attempting to work co-intelligently, a little background context. 

I was struggling with how to articulate working in a way where AI could best complement the work we do in HR. Then along came co-intelligence with AI and a breakthrough moment. 

Instead of thinking of AI as another ‘thing’ we must use in our work, it’s giving its best outcomes when you work with it collaboratively. It’s a partner to you, perhaps even consider it a co-worker in our day-to-day. 

Thing is, as I’ve experimented with co-intelligence, I’ve had a couple of important realisations and practical experience has really shown the reality of using AI. 

First, as obvious as it sounds, despite the way it might interact with you, AI is not human. It absolutely lacks self-awareness, empathy, conscientiousness and often will struggle to be organically creative and emotionally coherent.  

For everything it lacks, because it’s designed to please you it will, as recent news stories have shown us, try to find ways to work around challenges, sometimes resulting in it going awry or even lying. So, whatever AI produces for you must be human-checked and potentially edited. A phrase I heard recently that resonated with me was, ‘you need to aggressively micromanage it’. 

Secondly, it takes time to get great things from AI. Co-intelligence with AI doesn’t just happen; it’s a process of test and learn, back and forth. It’s not unlike when you start working in a new team, or with a colleague for the first time. With AI there are periods of adjustment too. For instance, working out the quirks, the balance between you, or even determining the division of efforts to best suit tasks you’re working on together. 

When you have cracked the division of labour and quality control your AI agent really can help. For instance, it has rapidly iterated and helped me develop my content in a much more rigorous way. AI’s capacity for research for example is also deeper, broader, faster and more thorough. With direction from me, its human colleague, it’s discovered powerful insights that I don’t think I could have unearthed alone in the time I had and depth to which I could delve. 

At the start I thought AI would just speed things up, like providing 10 basic text descriptions as a start point. However, with the right management by me, it surprised me by doing much more. An early piece of work, produced ideas and outcomes I hadn’t considered, but when drilling into them, quickly realised they would help shape our whole approach for the better. Then by building on the suggestions, working back and forth between us, the output continuously improved. 

This led to my AHAH moment (shoutout to Hive Learning)

• AI-assisted   
• Human-led   
• AI-resourced   
• Human-checked 

If we partner with AI like this, our skills and our work benefit from AI. Thank you, co-intelligence. 

In fact, after all the great experiments and awesome results, we liked co-intelligence so much, we made it part of PXO Culture’s theme for 2025, ‘resilience through co-intelligence’. 

 ‘Key things to takeaway’ 

1. Begin your journey with AI by addressing how to set yourselves up to work co-intelligently. This might include thinking about shifting mindsets, some skills development, behaviour change, culture evolution, policy etc. It’s all stuff that we’re familiar with and that we know must be done if you’re going to successfully adopt any new type of tool or tech. 
2. Bring out your inner-Centaur. Elevate the things we as humans know are vital to effective HR products and what we bring to the table. This enables our AI colleague to lean into what they’re good at – analysing data, producing in-depth reports, testing a hypothesis, creating gorgeous looking visuals, coding etc. 
3. Reject the narrative that to keep our jobs we must be instant AI expert users and the hype that AI is the holy grail we’ve always sought. On both counts, this is just not the case. What I would embrace instead, is taking the time to become a co-intelligent team member because it will benefit your long-term career development by ensuring utility of skill and relevancy of knowledge. 
4. This also isn’t an instruction to change ourselves. Instead, AI offers the opportunity to expand what we have, adding AI to your T-Shape for example, as a general capability offers the bonus of helping to enhance other skills.   

To close, some guidelines; 

• Always invite AI to the table, but remember it’s sitting alongside you, it’s not a VIP guest to be treated like royalty nor a slave to dump the crappy tasks on. It’s a paid member of the team, there to work and like you, have its strengths to be respected. 
• Always keep the human in the loop, especially when working in real time. Aggressive micromanagement might sound a little much, but it does need you to direct it to produce great results. 
• Always use a paid for AI tool. It makes for a more capable AI model to ensure best results. But never consider it to be perfect and especially not as the only one to use. Remember Mollick’s caution that the AI we’re using now is the worst it will ever be…so better is always available. 

Final thoughts 

From clients to colleagues, all of us are on different journeys with this, especially because implementing and using AI isn’t simple. So, noting the spirit of our community, ultimately let’s be human about this change.