My journey into gaming started with the introduction of the NES console into the house. My dad brought home the system and hooked it up in the bedroom, with the standard starter kit of Super Mario Bros./Duck Hunt, one of the laser guns, and the golden cartridge version of The Legend of Zelda.
One look and I was hooked - winking like a jewel I knew whatever it was, I wanted to do it. And so began my relationship with video games. I was six years old then, sneaking into the room to play as my then infant sister would stare googly eyed at the screen. This year, I finally got around to playing Breath of the Wild, accompanied by my now six year old son. It was a full circle moment in many ways - one, realizing I have been playing some version of Link's quest for more than 30 years, and two, that this obviously wasn't a phase.
I've written about my own journey into games for ESPN:The Magazine in a piece called "The Legendary Adventures of a Fearless Girl Gamer."
I've also made a doc series when I was at Fusion called Girl Gamers in response to what was happening at the time. It's my first documentary and I made a lot of mistakes, but I'm still proud of it, especially as I look at my growth.
A major flashpoint for the race and video games debates was the release of the trailer for Resident Evil 5. A hotly anticipated release back in 2009, Capcom's planned promotion went awry when the reaction to the images of black death sparked an international conversation about race and representation.
Capcom's PR teams were baffled - to them, they were celebrating a major coup in technical development. RE5 was the first survival horror game to incoporate sunlight as a major element of play, and they selected scenes to highlight the contrast between the sun-drenched streets and darkened doors and alleyways.
What never surfaced was the use of well-trodden imagery to showcase a white man (who in the trailer narrates his own antipathy for the mission) mowing through rows of black bodies, animated with superhuman speed and dexterity. In stark contrast to the ambling gait of the zombies in previous installments, the images were uncomfortable to anyone familar with the depiction of black bodies in popular culture.
To add insult to injury was the introduction of the female lead, Sheva Alomar. Introduced buttocks first, her fair skin and British accent immediately marked her as different in all the usual tropes, while ensuring that any nod to cultural difference would only exist for titilation.
This was far from the first time and far from the last time that racist imagery would appear in games. It can be traced back to the sexist/racist/settler affirming Custer's Revenge (1982) where the player reward is to rape a Native American woman. And it continues to today, where Overwatch removed a character's special "spray" (think finishing special effect) in the aftermath of the #BlackLivesMatter civil uprising. The spray? A hangman's noose.(There is also an interesting conversation to be had here around colonialism in games, particularly civilization games, but that's another discussion.)
But this situation was different - for one, the reaction of the gaming fanbase exposed a lot of racism and brought many writers of color to the forefront of the conversation to explain how their lives had been impacted at the intersection of gaming and race. It was also one of the first large scale campaigns attacking women critically responding to video games. Years before the term "Gamergate" would emerge in the news cycle, a coordinated group of gamers ran a blogger off the internet after a coordinated series of attacks. But that incident didn't receive widespread recognition - Sokari Ekine, the blogger behind culture and styleblog Black Looks, dealt with the attack, re-emerged quietly, and continued her original mission.
To this day, the issue of race in video games surfaces sporadically only to disappear as quickly. Part of the issue is the speedy censure for anyone who speaks out against racism in games. Another factor is the sheer numbers of people who can make change within the industry. The International Game Developers Association's diversity report normally shows women being roughly 20% of the industry as a whole. But the picture for blacks is stark, normally hovering between 2-3% of the industry inclusive of Europe, Africa, Asia, and the Carribbean. This dynamic is even present in esports, a subset of games that are played competitively, where even with the widespread participation in gaming
there are very few players of color that make it to the pros. The abysmal state of diversity in games still isn't a major topic of conversation.
I gave a little insight to my AI Background during Neta's talk, but to give a brief overview of how I got here...
I used to work at ESPN - while I was there, my tech fairy godmother introduced me to the Senior Executive Women in Tech group at Disney. I learned from some of the most amazing people in the business and was referred to the Disney working group for AR/VR. The two women that ran that group liked my thinking and asked if I wanted to join a new working group on AI and Machine Learning. I agreed and was introduced into a a new world. (A mentor also gave me a rough shove - he said that there was no money in AR/VR and if he was in my shoes, he'd try to learn AI.)
So I threw myself into learning everything I could about AI
and how it could apply to the business I was in, piloting some projects with the amazing folks who run Disney R & D, LucasArts, and ILM. But I wanted to launch more projects and faster. Also, I hit a pretty big block - you needed to learn Python to really advance in machine learning. We started a little working group at ESPN pooling the knowledge of friends who knew more coding and R, but progress was still slow. So I left. I left and didn't think I could reasonably do anything else in AI/ML - what could I contribute, if I didn't code and didn't have a PhD?
Luckily for me, I met Rediet Abebe, one of the co-founders of Black in AI at the MoMA and ran up to her. "Are you going to join?" she asked, immediately dismissing any of my imposter syndrome along the way. Through being part of the Black in AI community, I presented my work, a project called "AI in the Trap," at one of the most presitgious AI conferences there is.
Walking in the snow trying to print out my first ever academic poster was an experience - so was joining in on the coversations at NeurIPS that showed the major conversations happening around artificial intelligence and machine learning.
(If you really want to challenge your perception of reality, this was my favorite talk on body knowledge. It's a little grotesque -and not PeTA approved! - but with fascinating implications.) But I was in an encouraging community of scholars and activists trying to change the future of AI. And while an engineering background and heavy PhD work is what is normally focused on (for good reason, this stuff is hard!), a lot of the biggest innovations while come from people outside of the field trying to expand and solve problems with the same technology.
We are all needed here, because the future shouldn't be determined by a small group of people designing from their limited experience.
In the same way that Digital Love Languages asked us to imagine a world where the internet was designed differently, AI also needs that same thinking.
So my final love note is to share some of the things I learned from in hopes that someone will be inspired to take something they know and look at how AI is being applied there. The answers will suprise you.
And you can do it.
You're more capable than you think.
(At least, that's what I'll be telling myself while still trying to learn Markov chains!)
AI: a new (r)evolution or the new colonizer for Indigenous peoples?
Dr. HÄmi Whaanga
This is an essay by linguist and te reo MÄori specialist Dr. HÄmi Whaanga (NgÄti Kahungunu, NgÄi Tahu, NgÄti Mamoe, Waitaha). Dr. Whaanga warns of the potential for AI and related technologies to be used against Indigenous peoples as an extension of colonial practices of exploitation, extraction and control, particularly those that displace a peoplesâ understanding of themselves with a worldview that favors the colonizer. He discusses issues of data sovereignty in a technological landscape populated by AI systems existentially dependent on sucking up vast amounts of data on human activity, thereby putting Indigenous traditional knowledge and customary practices at risk of global-scale appropriation. Dr. Whaanga finishes his essay with a call to centralize Indigenous concerns in the work of establishing global ethical guidelines for the design and deployment of AI.
As IBMâs researchers thought about the challenge of making software follow ethical guidelines, they decided to conduct an experiment on a basic level as a project for some summer interns. What if you tried to get AI to play Pac-Man without eating ghostsânot by declaring that to be the explicit goal, but by feeding it data from games played by humans who played with that strategy? That training would be part of a special sauce that also included the softwareâs unconstrained, self-taught game-play techniques, giving it a playing style influenced by both human and purely synthetic intelligence. Stepping through this exercise, IBMâs researchers figured, might provide insights that would prove useful in weightier applications of AI.
IBM chose Pac-Man as its tapestry for this experiment partly out of expedience. The University of California, Berkeley has created code for an instrumented version of Pac-Man designed for AI studies; the company was able to adapt this existing framework for its purposes. (Teaching AI to play Ms. Pac-Man is a separate science unto itself, and a more imposing challenge, given the gameâs greater complexity.)
The researchers built a piece of software that could balance the AIâs ratio of self-devised, aggressive game play to human-influenced ghost avoidance, and tried different settings to see how they affected its overall approach to the game. By doing so, they found a tipping pointâthe setting at which Pac-Man went from seriously chowing down on ghosts to largely avoiding them.
This is a real problem with real impacts on peopleâs lives. Sure, a few incorrect Youtube captions arenât a matter of life and death. But some of these applications have a lot higher stakes. Take the medical dictation software study. The fact that men enjoy better performance than women with these technologies means that itâs harder for women to do their jobs. Even if it only takes a second to correct an error, those seconds add up over the days and weeks to a major time sink, time your male colleagues arenât wasting messing with technology. And thatâs not even touching on the safety implications of voice recognition in cars. So where is this imbalance coming from? First, let me make one thing clear: the problem is not with how women talk. The suggestion that, for example, âwomen could be taught to speak louder, and direct their voices towards the microphoneâ is ridiculous. In fact, women use speech strategies that should make it easier for voice recognition technology to work on womenâs voices. Women tend to be more intelligible (for people without high-frequency hearing loss), and to talk slightly more slowly. In general, women also favor more standard forms and make less use of stigmatized variants. Womenâs vowels, in particular, lend themselves to classification: women produce longer vowels which are more distinct from each other than menâs are. (Edit 7/28/2016: I have since found two papers by Sharon Goldwater, Dan Jurafsky and Christopher D. Manning where they found better performance for women than menâdue to the above factors and different rates of filler words like âumâ and âuhâ.)This is super geeky but important - even when people aren't trying to be biased the data approach might be flawed:
Automatic systems are beneficial to society but as they improve in predictive performance, comparable to human capabilities, they could perpetuate inappropriate human biases. In the context of sentiment analysis (SA), these biases may come in many forms. For instance, an SA system may consider the messages conveyed by a specific gender or race to be less positive simply because of their race and gender. Therefore, even when it is not clear how biases are manifested, it is the duty and responsibility of the natural language processing (NLP) system developer to address this problem.One quick final note. The military has influenced a lot of technology through funding and development and AI is no exception. It would be good to watch DARPA's perspective on AI particularly as they are such a major influence and funder in the space: