Antonym: The Three Slot Mind Edition

We're learning about ourselves and AI in weird new ways.

Dear Reader

We know multitasking is a misleading myth, but we still expect our brains to behave a little too much like a computer. Your prefrontal cortex, that executive function you think of as "you", has just three working memory slots. Not the seven-to-ten we once believed. Three. Try juggling more, and you'll drop something, feel that familiar cognitive discomfort creeping in –the urge to fight or flee from your own task list. A species of slow-build panic sets in if we’re trying to keep more than three things in mind at the same time.

Meanwhile, AI systems process thousands of variables simultaneously, holding entire knowledge domains in synthetic attention without breaking a digital sweat.

This mismatch isn't a bug, it's the defining feature of our new collaborative relationship in these thinking systems. We bring context, wisdom, and the crucial ability to ask "should we?" while AI brings the capacity to hold and manipulate complexity that would otherwise overwhelm our limited working memory.

Sometimes the overload comes from just thinking about what to do with AI

One executive I've been coaching, let's call her Helen, began with ChatGPT handling routine correspondence. Six weeks on, she's questioning how her organisation operates fundamentally. Why day-long meetings? Why white papers with ten PDFs sent across regions? What if the entire operating model is about to change?

This isn't unique to Helen. It reflects the natural arc of AI literacy development we described in Prepared Minds.

Pre-literacy → Foundational (faster tasks) → Intermediate (reimagined processes) → Advanced (new business models)

The Evolution of Meeting Data

Let’s take a look at a simple task evolves into a challenge to established systems as someone rises through the literacy levels.

Consider the humble meeting transcript. Once merely a record of who said what, it now represents one of the richest veins of organisational intelligence we possess, if we know how to mine it.

The progression from task to transformation follows a predictable arc, each level revealing new depths of insight:

Level 1: Task Automation (Do what we do better)

At the foundational level, AI transforms the tedious task of note-taking. Minutes and actions emerge automatically, saving 30 - 60 minutes per meeting (one EA told me it takes 1 - 2 days to write up some board meetings at their organisation).

We might stop here. Many do. But curious minds developing their AI literacy realise something: the transcript is data. Rich, highly useful data. You just have to ask some questions beyond who said what and what were the actions….

Level 2: Process Optimisation (Do what we do in better ways)

With new prompts, that same transcript reveals how discussions unfold. We can look at the meeting through the lens of actions, ideas, insights, psychological dynamics and performance (how well did we do in achieving objectives).

Thinking of it like sports performance meeting data can expose:

• Power dynamics: Who interrupts? Who gets heard?

• Participation patterns: Whose voices dominate? Who stays silent?

• Decision velocity: How long from problem to resolution?

• Energy flows: Where does momentum build or stall?

Helen realised that thinking of meetings as a team sport and the transcripts as data might help them have better meetings.

One organisation found male attendees dominated regular "collaborative" meetings, speaking over 75% of the time even when in the minority. This is not surprising, but the issue was more easily and deeply discussed when the data showed the imbalance, and made it easier for better chairing of the meetings and for attendees to be aware of their “airtime” as a measurable issue.

Another analysis discovered decisions were made in the first 15 minutes, with the rest spent rubber-stamping.

Level 3: System Transformation (Do new things)

Now the vertigo deepens. When sales calls yield UX insights, leadership talk reveals blind spots, and project failures are predicted by communication patterns, the very premise of meetings is in question.

Helen reached this level with her team's weekly strategy meetings:

• 15-minute async video updates per region

• AI synthesis of key themes

• 30-minute live decision-focused discussion

• Automated follow-through tracking

The time saved was incidental. The real insight? Meetings are often information-transfer rituals that AI can now handle, freeing humans for ambiguity, trust, and values-based decision-making.

At Brilliant Noise, we’ve even found that better prep and deciding who needs to be in a meeting has led to rich conversations that, when turned into data, get us most of the way to outputs like plans and proposals that would have been the actions. The meeting does the work, with some assistance from AI.

AI as a mirror

The most profound impact isn't efficiency: it's self-awareness. When teams see their communication patterns visualised, when leaders confront their domination of airtime, when organisations realise their stated values don't match their discussion dynamics, change becomes inevitable.

Another organisation discovered their innovation meetings had become risk-mitigation sessions. The language analysis showed 80% defensive framing ("avoid," "prevent," "minimise") versus 20% opportunity framing ("create," "explore," "build").

The journey from meeting transcript to organisational intelligence exemplifies the true power of AI literacy. It's less about tech, more about seeing familiar activities through new lenses. Every meeting generates data. That data reveals processes. Those processes encode culture. That culture drives outcomes.

When we learn to read these patterns, we don't just get better meetings. We get better organisations. The question isn't whether to analyse your meetings. The question is: are you ready to see what they reveal?

Show-Tell-Transform

I mentioned this last week, but I’m saying it again, since we’re on the subject of meetings: Were I to mandate one practice in every organisation trying to adapt to AI, it would be this: weekly AI show-and-tell sessions. Not presentations. Not training. Just five minutes of "here's what I tried this week with AI."

The format is simple: five minutes per person, no slides, just demos. One colleague might show a fact-checking GPT. Another, how they brainstorm strategy through voice chats with ChatGPT. Another, how they turn meeting transcripts into UX research.

Each small insight compounds. Knowledge flows. Most importantly, it makes not-knowing and learning through failure (everyone admits to their blind spots and where things didn’t work).

That’s all for this week

I hope this was interesting. These are complex times and we’re all thinking it through together. I’d truly appreciate any feedback on my attempt to explain what we’re learning and sharing. Reply to this email or leave a comment if you can.

Antony