Artificial Intelligence is Breaking Bad – But It’s Not Too Late to Fix (AI Ethics, Part 1)
The truth is, artificial intelligence (AI) is disrupting everything. It’s shaking up global labor markets and a wide range of industries from manufacturing, and health care to transportation and retail.
Digital giants are betting everything on it. Government agencies are pouring billions into it. And calls to prioritize ethical considerations in the use of AI are percolating through C-suites and boards everywhere.
The big question is: Should we trust AI to be fair in making decisions that affect our lives?
Joanna Bryson (@j2bryson)—ranked by Cognilytica as a Top 50 AI Influencer to Follow on Twitter—recently sat down with Appian and dropped some serious knowledge on the challenge of ethical AI:
- The consequences of AI bias in business and society
- How to root out bias in AI
- The danger of ignoring due diligence in software development
- Why regulation won’t take the magic juice (deep learning) out of AI innovation
- Why understanding the truth and consequences of AI is so important
Truth or Consequences
Bryson warns against a “move fast and break things” approach to mainstreaming AI. She also worries that we’re not prioritizing due diligence in the software development process. Case in point: The AI used in an automated system that suddenly cut disability benefits for thousands of people in Idaho.
Unfortunately, the system used bad data and a flawed formula that arbitrarily determined how benefits were allocated.
News reports said that the officials involved refused to reveal how the AI worked (under the guise of intellectual property).
The lack of transparency stoked constituent outrage. Which led to a class action lawsuit, which resulted in a court ruling, which forced agency officials to overhaul the flawed system.
Court finds #AI software so bad it's unconstitutional. Great work @aclu! Note that the software was judged by TESTING whether its outputs made sense, not by UNDERSTANDING its crappy code. Same can be done with ML/DNN.#aiethics #accountability #regulationhttps://t.co/kfiIXodTWB
— Joanna J Bryson (@j2bryson) September 2, 2018
Here’s the thing. AI touches virtually every aspect of our lives.
Smart systems make billions of decisions every day.
They determine everything from who gets soap from a soap dispenser, to who makes parole, to who gets approved for a loan, to who gets hired or fired from a job and more.
But what’s important, says Bryson, is for us to understand how the AI makes decisions, how it determines what we see and what we don’t, and who’s accountable when it goes sideways?