In this issue...
You may have noticed something different 😃 I have, after a long reflection, decided to refresh Quantum's template. I haven't made major changes to the structure, other than a few cosmetic choices, with the goal of making it more readable and slightly less amateurish. I hope you like it. As usual, my inbox is open, feel free to send a message 😃
·
Quantum #666 had an open rate of 50% and a click rate of 14%. The most clicked link was... the Wikipedia page about the number 666. You're an odd bunch, sometimes. That's why I love you all.
·
I had a great workshop at London's Design Museum, as part of a series for HMRC's senior leaders. The workshop was focussed on innovation, and it involved an interesting exercise where we were encouraged to see the section of the museum that presents technology evolution, and select an item that resonated with our view of innovation and explain why.
As I can never stick to the instructions, I selected two objects: an AK-47 assault rifle, and the famous One Laptop Per Child.
Looking at them in comparison and close to one another made me think about what drives innovation and how to deliver it well.
The AK-47 is a perfectly designed object: it's effective at what it does, it's got great usability, and its design is sound functionally. Apparently it's close to impossible to break it, and it works very well in sandy or wet conditions. It excels at good design, and its design is grounded in the reality of those who're meant to use it (these days, mostly irregular forces and insurgents, who may be working in very tough conditions). Obviously, it doesn't escape my attention that this is an item designed to kill human beings and, as such, deeply troubling from an ethical point of view.
What to say about the One Laptop For Every Child? It was conceived with a fantastic idea in mind: transforming education for children in the developing world so that they could become more self-sufficient in finding content via the laptop, with software specifically designed for it. Its prototype also came with a brilliant innovation: a hand-crank charger, and it was beautifully design. Where it failed was at its contact with the real world, maybe because of lack of contact between the West-based designed and the intended users: the hand-crank charger was not viable to really charge a battery, and with lack of infrastructure in the target communities, it became impossible to use.
So what do good products look like?
- they embed ethical thinking in their design, and they only exist for reasons that serve communities without damaging others
- they are designed well, with user needs in mind, together with the intended users, and those who might be subject to their use
- they are grounded in a deep understanding of the reality around them.
And this, of course, is a subtle metaphor for how to do AI well today, avoiding the hype: trying to solve real problems, with real users, understanding the context around them.
·
Regarding the previous discussion on tokens, I found something tangential but useful on Dense Discovery (issue 397):
The old rule of software was simple: as you scale, cost goes down, profit goes up:
“You write the code once. The ten thousandth customer costs you almost nothing to serve. Marginal cost trends toward zero, margins trend toward heaven, and the spreadsheet curves apart into that beautiful pair of diverging lines that venture capitalists have tattooed on their hearts.”
But AI doesn’t do this. Every query burns real electricity, real water, real chips. Revenue and cost haven’t diverged, they hold hands.
This is true and the biggest financial and cultural challenge, because most of us (I include myself) still think of AI as "another IT system". It is, but it's also very much not operating as one. Obviously, desktop software still used electricity, and cloud-based software did too. But the scale is so different with AI that it can no longer be considered "just another IT system". And I fear the lack of appreciation (see Uber's recent running out of AI budget).
·
On October, 21, I'll be speaking at Think AI for Government. Join if you're in London!
667 is a lazy caterer's number.
And now for this week's links.
