Suggestions
Sarah Gold
Designer, CEO at IF
Sarah Gold is a seasoned professional with a diverse background in the tech and innovation sector, currently serving as the Chief Executive Officer at Projects By If. With her prior experience as an Advisory Board Member at Techforgood and as a Co-Founder and Digital Service Designer at Wikihouse, she brings a wealth of expertise to her current role.
Highlights
🗳 Not unexpected, but nonetheless big news yesterday from #10. ICYMI here's what me, Rachel Coldicutt and Dr. Natalie Byrom think the next government should do to deliver better, cheaper, more accessible public services.
There are 6 recommendations on what must happen now and next in order to make the future possible:
👉 limit contract values with suppliers 👉 inspire a lost generation of technologists 👉 expand apprenticeships and in-job training opportunities 👉 invest in 12 regional and national Offices of Technology Innovation 👉 update our infrastructure 👉 build robust infrastructure for trust
People must be at the heart of this transformation. The focus must be on delivering real benefits, changing lives, and creating new opportunities.
Link to the work in the comments. And if you've not already, now's the time to register to vote!
“How many people realise the answer is wrong?...At what cost to them and to you?”
IMO people need to understand the limitations of an AI system.
I read Tom Loosemore's latest blog post and got thinking. Tom frames the limitations of AI systems in the context of service failure and cost. It's really smart. What does failure in systems enabled by AI cost users? What does it cost an organisation?
As I read the post I was saying to myself, “yes... AND we can and should make the limitations of AI more legible and understandable.” Because AI will always have inaccuracies. Systems always fail one way or another. So how do we design that as feature, not a bug?
Here’s one idea:
→ Apply a “Show confidence” pattern.
It's one of the patterns from our design pattern catalogue.
We’re currently building it for a client. Using context specific UI to surface confidence levels in a suggestion made by an AI customer support system. Then, we’re wrapping this in a service that supports people to understand the implications of different levels of confidence. And seek re-dress if the system is less than fully confident.
Note, it’s not straightforward, because… → confidence is subjective → confidence varies
But, by giving people the tools to make a realistic assessment of the system’s capabilities... → The system is more explainable and transparent → People can recalibrate their own expectations → The cost of (inevitable) failure states reduces
What’s not to love?
If you’re interested in design patterns for AI and how to implement them, DM me.
If this wasn’t enough of an encouragement to go read Tom’s blog post - Go read Tom’s blog post. Linked in the comments.
