Alex Roberts, AIDE
So you’ve been appointed to be your agency’s CAIO – what might you want to read over the long weekend?
Under the APS AI Plan(Opens in a new tab/window) every agency needs to appoint a Chief AI Officer (CAIO). (In December we in the AIDE team published some guidance about what agencies might want to consider in their choice.)
It’s a big task and an exciting one. But we all know that generative AI is a fast-moving technology that’s raising lots of new questions, and that means there’s no nice and precise job description or recipe for how to be a CAIO. So we in AIDE (AI Delivery and Enablement) want to help give the CAIOs as good a foundation as we can when they commence their new roles.
One element we thought might help is to put together a bit of a ‘starter pack’ with some resources that might help CAIOs as they start what may be a challenging role (leading transformation is rarely easy!).
And so I’ve put together the following with some picks from us in the AIDE team that we hope might help the CAIOs in their thinking, whether by challenging, inspiring or reminding them about what’s important. These pieces are chosen for their ability to stimulate rather than being chosen for being perfect or safe.
- Anthropic’s Andy Jones piece, “Horses(Opens in a new tab/window)”, giving a nice exploration of exponential shifts.
- Andrew Sorota’s “Rescuing Democracy From the Quiet Rule of AI(Opens in a new tab/window)”, a powerful provocation and warning about prizing efficiency and ‘algocracy’ without recognising the value and importance of friction and deliberation.
- Jennifer Pahlka’s “AI meets the cascade of rigidity(Opens in a new tab/window)”, a ripping read and a considered challenge about the risks of focusing too much on safeguards.
- A memo from the CEO of Shopify on ‘Reflexive AI usage is now a baseline expectation at Shopify’(Opens in a new tab/window), setting out a fundamental shift for the company, effectively mandating the use of AI by all staff in recognition of the fact that “What we have learned so far is that using AI well is a skill that needs to be carefully learned by… using it a lot.”
- Alex Komoroske on “What Got Lost in the Optimization(Opens in a new tab/window)”, outlining the dangers of focusing on optimising over transforming.
- A look at JPMorgan Chase’s blueprint for integrating AI(Opens in a new tab/window), an example of one approach of deep integration.
- The OECD’s “Governing with Artificial Intelligence: The State of Play and Way Forward in Core Government Functions(Opens in a new tab/window)”, which gives a great overview of how AI is being used in public services around the world.
- Tiago C. Peixoto’s “How to Guarantee AI Failure: A Field Guide for the Well-Meaning Senior Official(Opens in a new tab/window)”, outlining some cautions for CAIOs about being overly cautious.
- This piece from Nicolas Davies on “Your AI Strategy is Making Everyone Louder, Not Better(Opens in a new tab/window)”, outlining that when technology meets social animals, the social dynamics and coordination of teams is what matters.
I also asked members of the AI community of practice(Opens in a new tab/window) and some insights from them include:
- From Dan we have “Speaking and getting to know what their internal teams are already doing, trying to do, failing to do or wanting to do with AI first.”
- From Graham we have “AI governance and enablement can't be the mechanism by which all of an agency's tech and data process and governance roadblocks get addressed but governing and enabling AI inevitably involves reviewing and assessing processes from cyber, privacy, procurement, software licensing, etc. How do we deliver really good AI governance and best practice AI enablement while avoiding time-consuming process re-designs?”
- From Tennessee we have “Find your experts. The APS has many pockets of expertise in the research, development and application of technology, often tucked away in niche teams who are would be very excited to connect upwards and work out a strong way forward. Go outside silos. Go outside the APS as well, but don't overlook the internals either.” And “Have some fun. Enthusiasm is infectious.”
- And one I love is the ol’ classic from author Frank Herbert, “Show me a completely smooth operation and I'll show you someone who's covering mistakes. Real boats rock.” (i.e. don’t expect everything to be neat and tidy when it comes to trying to change things).
Are there other things that we should provide them? Are there other resources, guidance documents, mental models, critical insights or other things that get to the heart of the transformation effort needed to successfully adjust to the new possibilities of generative AI? How else might we need to help equip CAIOs as they start their roles?
If you have suggestions for what we should include in a CAIO starter pack, I’d love to hear them.