What I’ve been unlearning

Trauma-Informed Design Reflections #19

Lately I’ve had a lot of independent ideas marinating in my head, and I think it’s time to put them on “paper”.
(Spoiler: I think the last one is the best.)

On safe spaces

For the past few months I’ve been thinking a lot about safety. I’ve written about it, talked about it with survivors, sought it in therapy, and experienced pockets of it alone or in community.

And during all that time I kept coming across the term safe space and its alternatives. It was back in 2022 when I first heard a podcast by Brené Brown which introduced me to the idea of brave spaces. I’ve been seeing this term more and more lately. More recently, someone shared a medium story talking about accountable spaces.

Most of these alternatives are building on the idea that we cannot guarantee safety in a space. To do that we would have to control the bahaviour of everyone in it. Of course that’s impossible. Furthermore, what helps someone feel safe might result in someone else feeling unsafe. So how can we ensure a subjective experience?

I see this criticism as valid. Safe spaces are indeed a promise we can’t always keep. But I also see deeper issues.

I’ve written before (here and here) that for many trauma survivors safety might not be possible. Sometimes pockets of safety are, and if a safe space can be that, then it is truly valuable. But this could take time. In those cases “safe” is an attribute that is earned and that is very subjective, not something that we can throw out before we even start being together.

In addition to that, the idea of a safe space can often act as an expectation towards survivors. “This is a safe space”, when said by someone with power, can imply that “if you are feeling unsafe you are the problem”. This both invalidates someone’s experience and isolates them. Paradoxically, I have found that I feel the safest when I am able to say to myself: “you are allowed to feel unsafe here.”

Let me be very clear, I do not advocate for an unsafe space. Safe spaces are beautiful, and healing, and full of value for survivors. They can disrupt the everyday reality of trauma and help us move forward in our journey of recovery. But, the expectation that a space will be safe is usually unrealistic and can even be harmful. Perhaps we need to let go of the label and acknowledge the nuances of being humans together.

We can’t ensure that the space will be safe. But, we do invite everyone to strive towards that.

You’re allowed to feel unsafe here and we’ll support you if you do.

We’ll hold each other accountable.

On empowerment

A few days ago I was part of a conversation that touched on empowerment. My friend beautifully explained how in equity and liberation spaces the word is often avoided, because it implies a power dynamic.

This article explains it better than I can:

“It’s controversial. Certainly, anyone who offers to empower women does so with the best intentions. But language matters. This word is an outdated construct, perhaps particularly in the development community; “to empower” can sound neo-colonial and condescending. It represents us, those with means, giving power to you, those with none.”
- Karen Sommer Shalett, Editor-in-Chief, Impact Magazine

After the conversation I couldn’t but reflect on the use of the word by SAMHSA’s trauma-informed principle: empowerment, voice, and choice. Most importantly though, I reflected on the way I was using the word when considering this principle.

As was perhaps expected, I observed the same issue my friend was pointing out. I have been implying that survivors lacked power and that my design would “give power to them”. I felt ashamed with my observation.

I’m still exploring alternatives. For now, I find myself leaning more and more towards Chayn’s trauma-informed principle. Chayn has introduced the principle of power sharing which I believe addresses power differentials better. It’s not just the word that is changing, it’s a different way of seeing power and each other.

Hopefully I’ll explore that more in the future.

On AI (part I)

I have intentionally stayed away from writing about AI in this blog (with few exceptions). I believe that AI is both important and impactful for the trauma-informed design field. But so much of the AI conversation is currently the result of hype, that I’d rather wait for things to settle a bit.

This being said, there are two images which have fundamentally changed how I’m looking at AI. I’d like to share them here.

First, is a short blog-post by James Betker, a research engineer at Open AI:

Reading that for the first time a few months ago blew my mind. There are a lot of conclusions that come from this observation, but take them with a grain of salt. I’m not an AI engineer, and predicting the future of AI is extremely difficult.

The first is that given time, all the AI models will converge to the same points. In a few years the choice of model will probably be a matter of taste, similar to how the choice of a smartphone operating system (Android, iOS) is today.

The second is that since we only have a finite amount of quality data, there will probably be a limit to AI’s growth. At least with the technologies currently used. The lack of new data to train AI systems is already becoming an issue. It’d be interesting to see how much these models will improve in the coming years.

And the third is that data is extremely valuable nowadays. Because of the AI hype (and the real AI opportunities) a lot of funding goes towards building AI models. Because an AI model is primarily influenced by its dataset, better data will result in a better model. So it makes sense that most companies are doing their best to collect more user data, since this can potentially result in a better AI model. But, it is worth considering here that more data is not the same as better data.

On AI (part II)

And then there’s this comic from theforestjar:

As with any art there are multiple ways to see this. The one I’ll touch on is this: “AI is not the problem, the systems of oppression inside of which AI is being developed and used are the problem.”

There are many issues with the rapid growth of AI. The loss of jobs (especially in copywriting and customer support), the exploitation of annotation workers, the rise of pornographic deepfakes, and the spread of misinformation are a but a few of them.

And even though AI is connected to those problems, it would be a mistake to view them outside of the context in which AI operates. After all AI development happens in systems of capitalism, white supremacy, and patriarchy, all of which exploit people and exercise power over.

So when AI results in putting profit over people, exploiting the Global South, and abusing women, how can we blame the tool and not the master?

In tech, we often see those issues as “unavoidable collateral damage”. And they might be unavoidable inside those systems, but it’s worth questioning:

Is there no other way?