Thread by @npscience

Prof Ifeoma Akjunwa @iajunwa @CornellEng up now to discuss not doing harm with algorithms #sageassembly #techandethics

She asks: Do we need a “do no harm” principle for tech?

A reminder that algorithms are old, they’re just maths and rules.

What has changed since the abacus ring? We retain the data in today’s world. Who owns, controls and uses these data trails?

#SageAssembly #techandethics

Of note, these data include health, location, interests, dating profiles, political persuasions...

All very well saying AI could be used for good. Cambridge Analytica reminds us to check our trust in algorithms and the people applying them.

#SageAssembly #techandethics

(^^ All these tweets are rough quotes of @iajunwa)

: algorithms aren’t just numbers & rules: they affect our experience, they are written by humans who bring inbuilt biases, they are used to change our lives.

Algorithms right now are a black mirror, reflecting the weaknesses in our society.

#techandethics #SageAssembly

asks who is using the algorithm and for what? To improve individual experience?

What if companies could predict which women are about to get pregnant and fire them first?

#techandethics #SageAssembly

: algorithms have changed organisational behaviours, they have crept into hiring practises, from Walmart to Golden Sachs.

Note that automated hiring systems are culling systems. Efficient rejections. What are the rules they are using?

#SageAssembly #techandethics

: is it neutral to include a rule about gaps in employment? What about parents, carers, people who have spent time in prison? Who are these rules excluding?

#SageAssembly #techandethics

other rules include

* Pattern matching: this finds the same people you’ve already hired. Who don’t you hire already?
* Personality tests: or are they? We hear an example where this was v close to a mental health diagnostic test.

#SageAssembly #techandethics

: even if the hiring rules are solid, how would we know? Who is inspecting and regulating? Who is training these algorithms are what are their motivations?

#SageAssembly #techandethics

uses the example of a hoax chess program (under the hood was a real chess master) to highlight that algorithms are human: human-designed, built, trained, applied. We as humans need to reclaim the responsibility for algorithmic decision making.

#techandethics

asks: what if CEO stood for chief ethics officer? Isn’t ethics the most critical function of an org? It is required for consumer trust, ethical tech offers a route towards success.

Prof Ajunwa closes with an ask: move slow and fix things.

#SageAssembly #techandethics

Great Q from audience: consider the motivations of those who shout loudest when scandal breaks (eg Facebook).

@iajunwa: we‘ve been socialised to give away our data via gamified social platforms. This is where regulators need to step in. And we are undereducated about this.

Super interesting discussion after w/ @madprime: is it ethical to go slow to be ethical (and potentially not make anything or make too late) whilst others go fast and make products and outcomes faster (and thus get adoption)?