On My Mind - #3
Gender Fluidity, Gen Z, and AI to cheer you up.
5 things on my mind - by Michael Dempsey
Apologies for the gap in sending these. Back to our semi-regularly scheduled programming. This email spawns from this thread. The process for this will evolve but as you'll see, some thoughts are random, and most are unfiltered or poorly edited. Either way, let me know what you like, don't like, or want to dig into more.
5 Things
1) On the rise of gender fluidity and digital identity
I've written a lot about identity, but mostly related to avatars and how they can expand, fragment, and change digital identity. However, when talking to multiple founders in the avatar space, some have lamented about having to anchor the creation UX with a gender choice at the top of the funnel (i.e. choose "male" or "female" to begin). Memoji explicitly doesn't mention gender, allowing you to build your character outside of any implicit societal restrictions. They are one of the only I have seen make this decision.
What is really interesting to me is that these products, often built for Gen Z, could allow for a decreased weight to be placed on gender and how we think about our identities. The internet has always been a special place for self expression, but perhaps when we make users think about how they physically portray their identity digitally, we can capture what actually makes up an identity outside of the first checkbox of gender and second of race. A lot has been written about this, and I admittedly am not the right person to speak at length about this specific topic, but it is on my mind and I thought this article summarized a lot of the impending effects we could see via the rise of gender fluidity in the world over the next few years across both startups and corporates.
2) Parsing the noise of Gen Z
A few weeks ago I tweeted this (above). Based on a few DMs I wanted to elaborate on this thought.
First, I don't think being Gen Z or spending time there with your resident Gen Z advisor really gives you a massively valuable edge. You at best get a snapshot of behavior, but not something with perspective or that is well-informed about how these cohorts will progress in their behavior. The early usage of technology may have a few gravitational centers (self-obsession, for example), but I'm not sure it's a long-term knowledge advantage.
Second - Knowing what will be a fast hit may have been interesting as an investor yesterday but IMO in today's social climate that isn't interesting to me as a VC. My job isn't to invest in the next $50-$150M FB acquihire and because of often the profile of Gen Z focused founders, as well as the dominance of the social platforms, you need something and someone incredibly special to let these things run at a pace that makes the business look bulletproof enough that when facebook paints a target on your feature set, you'll survive.
That said, this cutthroat social climate also may mean that the past training data that I said I wish I had may be irrelevant in a segment where company dominance has never been so strong.
P.S. - I know about Zebra Intelligence. If you’re a brand trying to understand Gen Z today, use it. #sponconbutnotreally
3) We need more robot-first environments.
While many people (myself included) are excited about the potential for machine learning/computer vision to bring true scalable autonomy to robotics, in the near to mid-term I feel that multiple use-cases could benefit from a re-thinking of the problem. There are multiple interesting reddit threads and articles that discuss the issues of environmental compatibility with certain types of robots.
The most notorious issues are within self driving cars/autonomous vehicles where you can constrain the problem to robot-friendly environments. I previously called these Hedge Cases. Some argue that city infrastructure is causing issues with AVs, like this specific thread on the infamous unprotected left turn problem or more recently Ford discussing a future without traffic lights.
"This is a little like saying that computers will not be able to be used everywhere because not everywhere gets electricity. That's true but it isn't microsoft's problem...Right now SDC's are the ones that need to accommodate current infrastructure because there are so few of them. But in a couple of years it is infrastructure that will need to take them into account."
What’s more interesting to me is in the realm of consumer robotics. There's a lot to unpack here but my main view is that with the massive amounts of new real estate development happening everywhere, it feels like low hanging fruit to bring true automation into the house. Or as another person in this thread put it:
"We could build robot-friendly houses, just as we built dishwasher-friendly houses and laundry-friendly houses."
4) Sadly, at seed, it’s not always great if VCs feel they understand your entire business.
Still thinking through this…
5) AI to cheer you up
I randomly found this github repo and was surprised I haven't seen this experimented with more. The idea of setting a human emotion as a goal for a reinforcement learning model is kind of brilliant. This model essentially uses emojis to seek for the goal of a smile from the user. As dystopian as it is, I can't wait for my “cheer me up” AI. For now I'll just use Try Not To Smile.
A few papers/links this week
Presentation on the rise and fall of permissionless systems.
Audio2Body Dynamics - Taking audio from a piano or violin and modeling what the physical manifestation of that audio would be (i.e. how would someone look playing it).
Do GANs leave artificial fingerprints - A deeper dive into if, like cameras, GANs leave fingerprints that could forensically tie the media they generate back to a source. This is another step forward in detecting digitally manipulated assets.
If you have any thoughts or feedback feel free to DM me on twitter. All of my other writing can be found here.