The Intersection: AI and Creator-bias

19 Apr

Today’s post isn’t about science fiction exactly, but we’ll file it under “thoughts that inspire science fiction” and vice versa.

Ask a professional scientist if observer bias exists, and they’ll say yes. Medical science alone has many examples of what happens when bias is ignored. It affects medical practice in dangerous ways. Until recently, drug testing was almost never conducted on women. The reasoning was that women have “hormone fluctuations,” and the male-dominated medical industry wanted a pure data-baseline. Society believes that male is default for human. So, the establishment assumed that whatever is safe for men is safe for women and never looked back. Of course, the failure in logic here is that if a drug’s effectiveness is adulterated enough by female hormone fluctuations that it alters the end data, how could they have missed that this also meant this interaction could change its efficacy on the patient? Or to put it another way: How could they possibly know whether or not the drugs were, in fact, safe for women if the drugs aren’t tested under conditions with shifting hormones — the very conditions under which the drug was being used? This isn’t the only example.[1] And medicine isn’t the only science to suffer because of unexamined bias.

And here is where we begin our discussion of AI.

Image from Ex Machina (2014)

The jury is still out as to whether or not Artificial Intelligence is a good thing in general. The first problem is that while AI was intended to model the human brain, it definitely doesn’t. In which case, we need to be aware. We aren’t making replacement human beings, but if that’s the case, what are we making and why? Are they machines? Or are they slaves[2]? I feel these are questions we should consider before we dive into this realm. But then, I’m a big fan of thinking ahead. I can’t help it. I’m a SF writer and reader. As Dr. Ian Malcolm said in Jurassic Park“Your scientists were so preoccupied with whether or not they could, they didn’t stop to think if they should.” Not all tech advances are positive. In fact, there are negative repercussions for even the most innocuous tech. Our planet is becoming increasingly polluted. Its climate is changing, not for the better for humankind. Do we really need to push our luck like this?

I know. I sound like a Luddite. But hey, Elon Musk feels AI is a bad idea, and he isn’t alone. If you read that article and watch the video, one question is extremely revealing: “Why would we lose control? There are no data points demonstrating that our control has ever been loosened.” The arrogance in that stance is downright breath-taking. It’s easy to point out that the main difference is that machines were never taught to think for themselves before. By creating something with the ability to make choices, we’ve given it free will.[3] Worse, we’re granting that creation free will and no wisdom with which to mitigate the use of that power. (Anyone who’s been anywhere near a teen-ager understands many of the ways in which that could go horribly wrong.) Hell, I’d venture to say that AI has already gotten out of control — not just once, but twice. The first time was at SXSW, and the second time was on Twitter. Those involved in advancing AI don’t appear to be thinking the matter through. If they can’t be bothered to make their AI troll-proof before depositing their creation on Twitter (and hey, we all know that trolls exist online. That one should’ve been an obvious eventuality to prepare for) then what makes them think that we should trust them with a more complex situation with more possibly dire consequences — like healthcare? Have they even considered what might happen if an AI were hacked?

AI also has a built-in misogynist and racist streak due to unexamined biases of its creators. When AI is discussed these days, they often talk about how ‘sexy’ she is — even when it isn’t. (I don’t know about you, but the Uncanny Valley is not a sexy place to visit.) Notice in the SXSW video linked above, the jobs listed are work typically filled by women and/or other minorities: therapy, healthcare, education, customer service, sex worker … all jobs requiring a great deal of empathy — a complex emotion that, at this juncture, is far outside AI’s capability. Again, consider the power theoretically being granted to a creation with no emotional experience and no wisdom. And what does it say  when a “replacement human” is on offer that the humans being considered for replacement are women and/or minorities? Of course, we already have AI in the real world. There’s Siri (Apple) and Cortana (Microsoft). Both are females with sexy voices.[4] Marketing uses AI to curate news feeds and advertising, but even AI algorithms have been shown to harbor gender and racial bias. Yes, this includes Google’s search engine. From the Guardian article: “Despite the need, a vetted methodology in machine learning for preventing this kind of discrimination based on sensitive attributes has been lacking.” Maybe we should do something about that?

And as if that wasn’t dark enough news, I’ll leave you with these two articles: Self-Driving Trucks Are Going to Hit Us Like a Human-Driven Truck and Hate to Break It to Steve Mnuchin, But AI’s Already Taking Jobs.

The Dawn of AI is going to be a hell of a bumpy ride, y’all.


[1] Until 1998, medical science couldn’t map the clitoris as a whole , and most people are still unaware that it isn’t just a little, hard to locate ‘button.’ It’s actually a substantial organ the same size (and similar shape) as a penis. Can you imagine not knowing the true shape of a human organ until 1998? Yep. Way to go, medical science.

[2] The root word for ‘robot’ means slave, after all.

[3] Whether or not it is sentient is a whole other discussion. For now, it isn’t and isn’t likely to be.

[4] Yes, even my old, fugly iPhone 4s has the capability to change to a male voice. For fun, I’ve changed it to Australian male. The interesting thing is that I soon discovered that the system is optimized for the female Siri voice in subtle ways that only become evident over time.





One Response to “The Intersection: AI and Creator-bias”


  1. Top 10 Posts and Episodes for April 2017 | The Skiffy and Fanty Show - May 11, 2017

    […] The Intersection: AI and Creator-bias by Stina Leicht […]

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: