Mrs. Davis and the Certainty of the Algorithm

2023-06-24 ☼ culture ☼ technology ☼ blog

In Hello, World, Hannah Fry describes the “brilliant decision” of IBM engineers to make Deep Blue feign uncertainty by occasionally adding an unnecessary pause between completing its calculation and announcing its next move. These delays made it seem like the machine was struggling through difficult calculations, creating a false impression of fallibility that ultimately undermined Garry Kasporov’s judgement, and contributed to his defeat.

Uncertainty is valuable when the relationship with the user is antagonistic. When a system is meant to be offering help, on the other hand, absolute certainty is far more reassuring. This is what large-language models like ChatGPT project, confidently offering an answer to any query without acknowledging the possibility that it might have concocted a hallucination. This is one of the features that make them a particularly compelling source of information:

People form stronger, longer-lasting beliefs when they receive information from agents that they judge to be confident and knowledgeable, starting in early childhood […] People regularly communicate uncertainty through phrases such as “I think,” response delays, corrections, and speech disfluencies. By contrast, generative models unilaterally generate confident, fluent responses with no uncertainty representations nor the ability to communicate their absence. This lack of uncertainty signals in generative models could cause greater distortion compared with human inputs.

The recent Peacock series Mrs. Davis was written before the emergence of ChatGPT, but it recognises how our relationship with technology is defined by certainty. The titular Mrs. Davis (also known as “Madonna” in Italy and “Mum” in the UK) is an algorithm that effectively controls the world, orchestrating billions of tiny good deeds that are supposed to collectively tilt humanity in the right direction. Mrs. Davis offers absolute certainty in its guidance. No one needs to struggle with doubt any more: simply do what the algorithm tells you, and rest assured you are contributing to the greater good of the world.

Simone, the protagonist of Mrs. Davis, is a nun. But faith demands uncertainty, and it cannot exist in a world without doubt. At the heart of the show is the relationship between Simone and her mother, Celeste, a character defined by her inability to tolerate uncertainty (a trait that is reflected in her career as the head of global espionage agency, an organisation responsible for shining a light into all the shadows of the unknown). A climactic moment in the series hinges on Celeste’s insistence on knowing, despite Simone’s desperate attempts to convince her that sometimes ignorance is preferable.

The appearance of uncertainty made Deep Blue seem fallible and therefore human, as it excelled at a task previously believed to be the exclusive domain of human creatvity. The closing image of Mrs. Davis — a windmill turned not by wind but by a person pedalling — represents an inversion of Deep Blue, a human performing the function of a machine. Being human, Mrs. Davis proposes, means tolerating uncertainty, and trying to do the right thing even though we never truly know for sure that we aren’t just making things worse.

Edit: Updated 20230625 to include quotation from Celeste Kidd and Abeba Birhane