We do not yet fully understand the implications of artificial intelligence — particularly in professional services, where the disruption is already material. Nobody has all of the answers, and yet there is no shortage of people claiming to have solved for different parts of it.
The reason everything feels so unsettled is that we're all operating at different altitudes.
I’ve observed — and the evidence supports — that the less someone understands the complexity and nuance of large language models, the more bullish they tend to be.
At the opposite extreme, the leaders of the AI research labs — the Sam Altmans and Dario Amodeis — are at pains to warn us about existential risk; risks which, not coincidentally, tend to point toward the need for large-scale, ongoing investment.
And then there is the messy middle.
The thinkers, builders and operators trying to make these tools work in practice are often both elated by the possibilities and frustrated by inconsistencies and unreliability. Capability is uneven, outcomes are difficult to reproduce, and the gap between what a system can do and what it can be relied upon to do remains the more honest measure. Bridging that gap is harder work, but it leads to better outcomes.
The organisations getting it right are taking a slower, more deliberate approach to transformation - one that creates the space to understand both the tools and the underlying structural or business problems they’re being applied to. This stands in stark contrast to the prevailing sense of urgency, and the anxiety that comes with trying to keep up without the time or structure to do so.
And then you see a tweet from tech founder and VC Marc Andreessen claiming to have written a prompt that makes a large language model “10x smarter.”
That’s not a breakthrough. That’s a category error.
It reduces a complex, probabilistic system to a single input. It treats intelligence as something that can be dialled up through phrasing, rather than something bounded by architecture, training data and constraints. It confuses a better response with a fundamentally more capable system.
And yet, this is the altitude we amplify. Not because it’s right, but because it’s simple. Legible. Lazy.
Which leaves the rest of us trying to make consequential decisions in an environment dominated by ideas that flatten the very complexity we’re dealing with.
And leaves us with a question: in an era of such turbulence, who are we choosing to fly the plane?