Let’s be honest: for most workplaces diversity and inclusion is about checking a box. I do believe that people at the top of those places value diversity in principle, just not enough to overcome the inconvenience of change.
I’ve worked in tech for eighteen years. There was one place where my manager would hire women based on how hot they were. He did this with HR’s blessing. That department was 92% men but praised itself on diversity.
I have so many stories like that. My friends do too, including those that give trainings. DEI efforts are almost always performative.
Enter AI slop
Larger businesses and organisations will often have a portal with learning materials on DEI. These portals are typically bundled with another product the company needs like health insurance. The content was an afterthought composed by non-experts.
So of course, these companies are now embracing AI. I’m seeing an entire catalog of DEI learning materials be replaced with AI slop. No one involved in the transaction cares about the quality; only the subjects and experts would and they are excluded from the process.
Capitalism.
One example
I do want to talk though about the kind of bad content AI introduces.
Here’s an example from one platform. It’s the one page discussing how to address gender-based discrimination in the workplace.
Support Non-Discrimination & Anti-Harassment Initiatives
Advocate for non-discrimination and anti-harassment initiatives. These are essential in safeguarding
everyone’s rights and creating an environment where respect and unity are paramount.Learn About Gender Inclusivity
Participate in open discussions about gender inclusivity to build understanding. Your input can lead to meaningful changes and ensure that you and your colleagues feel included.
Challenge Your Own Gender Biases
Regularly reflect on and question your own beliefs and biases about gender roles and norms. This self-awareness can help you interact more inclusively with
colleagues.Support and Amplify Underrepresented Voices
In meetings and discussions, make a conscious effort to encourage and amplify voices that might be underrepresented.
Participate in Gender-Related Learning Opportunities
Attend workshops, seminars or webinars on gender inclusivity. This continuous learning can enhance your understanding and effectiveness in fostering an
inclusive environment.Engage in Constructive Dialogue
When you observe gender biases or stereotypical remarks, address them in a respectful and educational manner. Engaging in constructive dialogue can foster better understanding and change.
This may seem good at first glance. The devil’s in the details.
First, this doesn’t teach anything. You could distill its entirety to don’t discriminate. The problem with that is that no person who discriminates believes they’re doing so.
That’s why you need to name what behaviours are inappropriate. You also need to name who is victimized by them because you can’t effectively address problem if you ignore the most important predictor of that abuse. Men are disproportionately perpetrating gender-based workplace discrimination and they’re doing it against women. Start by addressing specifics there.
The AI content avoids this because it’s biased towards producing milquetoast content; pooling words that don’t elicit a reaction. In this society get offended when victims of discrimination speak up, so it is omitted from the AI model’s output.
Just like it doesn’t use the word “woman” once in this article about sexism.
A second example
There’s another page on “LGBTQ+”. Here’s how it opens:
LGBTQ+: Equal …
Don’t assume that being LGBTQ+ is the cause of any problem in the person’s life. However, due to discrimination, they may have to deal with particular stresses.
Do treat LGBTQ+ persons equally, as you would any other identity. Open events to all, and avoid assuming which individuals would or wouldn’t want to be invited based on stereotypes or assumptions.
… But Different
Acknowledge that the experience of LGBTQ+ in a homophobic and heterosexist society has a profound effect on how LGBTQ+ individuals view themselves and how they experience the world.
This page never defines the LGBTQ+ acronym and mixes in advanced terms like heterosexist without explanation. It’s entirely inappropriate for this rudimentary material. It also continues with the empty just don’t discriminate messaging.
None of this content makes sense from an educational perspective. But an AI algorithm isn’t trying to educate. It’s assembling a sequence of words based on probabilities seeded by the prompt.
The page continues:
Good Conversation
Don’t limit conversation with LGBTQ+ individuals to LGBTQ+ issues.
Talk about the same things you would talk about with anyone else – the weather, sports, hobbies, etc.
Do listen to and respect boundaries of LGBTQ+ people regarding what they do and don’t want to share or talk about.
Partners
Recognize same-sex relationships as equally legitimate as opposite-sex relationships.
They’re people too! This is feel-good slop that won’t move any needle anywhere. It continues:
(Partners, continued)
Use gender neutral and inclusive language, such as partner, significant other, and committed relationship. If inquiring about whether person has a significant other, use that term or other open, inclusive, gender neutral terms.
Listen to your friends as they introduce their partners and use the same language when introducing them.
Don’t use boyfriend, girlfriend, husband, wife or marriage, unless stated as their preference.
Transgender People
Don’t assume an individual’s sex or gender identity based upon their appearance.
Use trans-friendly language, such as “he, she or they,” or “people, employees, associates.”
Ask politely what pronouns and name a person prefers you use when referring to them.
Don’t use transgender as a noun or terms like “transgendered,” “transsexual,” or “transvestite.”
No expert given this number of words to work with would give this advice. There’s no explanation for why any one of these points might matter either making them trivial to disregard.
Missing is any recognition of systemic workplace barriers. It’s all framed as isolated interactions among peers. This goes back to AI-generated content being inherently inoffensive, it does not question workplace structure. But that’s the most important area to address discrimination: washroom access, names in HR and IT systems, gender-based benefits, reporting structures, etc. It’s also one everyone has a role in, including those outside of HR and IT.
AI content is worse than nothing
Back when DEI-related learning materials were written by copywrighters, I defended their content as being better than nothing. There was at least a chance that the freelancers would paraphrase something of value they found online.
This AI-generated material with its filter to make everything as inoffensive as possible, so that even perpetrators find it palatable, is so much worse.
The world would be better without it, but companies will keep paying for it, because doing the right thing would be inconvenient.