Brave Spaces

- How did an AI generated sitcom parody land its developers in hot water?

By sharing transphobic and homophobic jokes.

A "Seinfeld" parody continuously livestreamed on Twitch for 54 days.

The equivalent of watching upwards of 3,500 actual episodes before its offensive jokes resulted in a temporary ban and an editorial overhaul.

And while it's not uncommon to hear this type of rhetoric online, it's a little jarring to hear it coming from computers.

To understand how we ended up here, what it might mean for queer communities and who is looking out for us.

Today I'm sitting down with Leanne Pittsford, CEO and founder of Lesbians Who Tech & Allies, the largest community of LGBTQ+ technologists in the world.

- If we don't move and create urgency in the things that we want to see shift, no one else is going to do it for us.

(crowd cheering) - I'm Devin-Norelle, and this is "Brave Spaces."

- I started to connect the dots to who had power and who had access to power between things like race and gender and sexuality, right?

And so, realizing that and realizing who was making the decisions, I wanted to be part of changing that, right?

I wanted community, I wanted a seat at the table.

Growing up, I definitely had one of those, you know, a little bit of a "Lifetime Special" situation where I didn't totally know I was gay because I so desperately wanted to fit in in my community and my home and my family, right?

I, like, wanted their love.

- Right.

- You kind of learn and it's sad, right?

It's not always the best, like how to think outside the box in this way 'cause you're hiding.

I mean, we, like, burned our journals and, you know, we're always trying to, like, figure out how no one would find out.

And I think those experiences give you the skill of problem solving.

Like, so I would've been a perfect type of person to go into engineering, right?

Even entrepreneurship business.

I just got pushed, as a woman, away from that space.

Not much has changed.

You know, when you, especially when you look at technical teams, in terms of representation, and obviously, - You still see white men?

- Yes, there's nothing specific for our identity.

So I thought if we just intentionally focus, like, being inclusive without being exclusive, maybe we'll get people that want to participate.

- Queerness in Tech have a deep interconnected history.

From the early days of instant messaging to the rise of YouTube, Tumblr, TikTok and dating apps.

The internet has revolutionized the LGBTQ+ experience.

I had a mother who said I was not allowed to date, and I was not allowed to be around other queer people.

She actually banned me from going to the gay parade.

What I ended up doing is, taking to the internet around the time when AOL dial-up was a thing.

- Yes, AOL chats, AOL chat.

- I'd get online and I'd find the gay chat rooms, I'd make friends, and really that was my way of finding community.

- I mean, I really had not one person that I would even think that I could turn to to have a conversation.

I know in some way that's a huge part why, where I ended up working, you know, in LGBTQ rights, even before I started Lesbians Who Tech & Allies.

- [Devin] For so many, online spaces are safe harbors with open access to knowledge, community, activism, and the ability to live authentically.

It has transformed the way queer people come together.

This is especially important for young people, because around 60% report feeling like their home is not LGBTQ+ affirming.

The internet is also a place for extreme, hateful ideas to take root.

And because we're so online, we have unprecedented access to all of it.

The good, the bad, 24 hours a day, right at our fingertips.

Here's the truth, the tech world, the folks responsible for building and safekeeping our online spaces, skews predominantly male, white and heteronormative.

And this lack of representation extends far beyond our problematic bit of parody.

The demographic makeup of the tech industry can be discouraging for women.

By hosting programs and opportunities specifically catered to lesbians and their allies, Leanne and her friends are working to connect and empower women to pursue careers in tech.

- Well, I think going, even knowing that Lesbians Who Tech exist, I remember reading the name first and I was like, "What does that even mean, or why does that need to be?"

And I think in the name, I think those questions that arise are really important.

- I think both in terms of professionally being able to see out queer women who are doing things that, you know, just even how you think of your career, like as someone who does not come from wealth, like, first person family to graduate from college.

Like, there haven't been a lot of possibility models, and I feel like I've gotten a lot of that.

- People hire more people that look like them.

Now is that always true?

Absolutely not.

But the probability would predict that's true.

- Because tech holds so much power, it's exceptionally important that it become a representation of what is possible because it could bring us into a utopian moment and age.

That's, hopefully, what it is we are trying to do with it.

- [Mia] I will say, when it comes to AI, I completely agree.

I mean, it has the ability to actually get us to that utopia state.

I would be more concerned about having putting, like, having pause and regulating, is a thing that we should already be regulating, which is greed.

- Despite its recent exposure into the mainstream AI has been around for decades and can be found just about everywhere, from web searches to targeted ads and weather reports.

More complex systems such as those that generate content and images, or mimic human-like conversation, fall under the umbrella of generative AI.

Large language models or LLMs are one form of generative AI used to process and generate language.

LLMs can produce sentences that generally make sense, logically and grammatically.

And in order to mimic human thought and speech, they have been trained with data, essentially fed massive amounts of text using billions of words from the internet.

- All a large language model does, is that it predicts what's next in a sequence of words.

The unique thing, is that computing power has gotten so good and computationally efficient, the computational efficiency has allowed us now to take all the vast amounts of language we have from the internet and then use that to say, "Okay given this context, what do we predict as the next word?"

- So when you think of it, AI knows what we humans know, and therefore what we teach it from developers and computer scientists who are hands-on with the product, to everyday people publishing their thoughts and inadvertently sharing their data online.

- So when you have garbage in, you get garbage out.

And we have a lot of, unfortunately, heterosexists, you know, homophobic, racist, misogynistic garbage on the internet.

And that's the freely available data that are used to train these models.

And so we can't be, we should not be surprised, when they reflect the same biases that we wade through every day using the internet as people of marginalized identities.

- Who is going to make sure that we're protecting them, that we're keeping them safe?

Because this thing is just going to grow so fast so quickly.

- So, it is one thing for AI to make jokes that replicate human biases and discourse, but unchecked, it could be much more consequential, affecting queer folks at school, in the workplace and beyond.

- [Leanne] We spend so much time and energy on our phones, on our laptops, we communicate with each other through these avenues.

So obviously it holds this place in society that few other things do.

So, making it equitable, therefore, kind of reshapes everything that we're doing and how we're communicating to one another.

- Like, I think about two things.

One, we have to skill people, period.

We have to skill our community on this.

This is not a phase, this is the future.

And I know underrepresented people have lived their life in such a way they will be thinking about those things more clearly, and will have better ideas, frankly, to solve those problems because they've gone through it.

But if we don't have different types of people in this room, specifically with generative AI, it's going to get so much worse.

- Generative AI will soon be unavoidable.

So we need women and LGBTQ+ folks to feel welcome to tackle these challenges head on.

Sadly, half of all women end up leaving tech by the age of 35 due to a culture of exclusivity.

Shout out to the talented women providing mentorship and support for lesbians in tech, so that we can get ahead of the machines.

As anthropologist and media scholar, Mary Gray put it, "When AI whittles us down to one identity, we can look at that and say no, I'm more than that."

Thanks for watching this episode of "Brave Spaces."

I'll see you next time.

- A lot of what I think about is this story about this woman and she's just like chipping away at the foundation of this building and, like, days go by, people walk by her, but like, because she just keeps chipping away, eventually the building will fall.

And so, like, it won't happen every day, but, like, we have to create the enabling environment for the next generation to do that.

Like, that's what gets me up.

(upbeat music)