different worlds, inverse scaling & what it means to be human
CC#62 - An (Eu)Catastrophe, Placebo Effects for Creativity and Thoughts on Nuclear Power
Hey there and welcome to ✨ CuratedCuriosity - a bi-weekly newsletter delivering inspiration from all over the internet to the notoriously curious.
Things I Enjoyed Reading.
I feel like a lot of the conversations around AI are very binary these days - either they paint a future were humanity is doomed or some amazing utopia where no one needs to work anymore. Both narratives don’t really give any room for human agency - this article, in contrast, points out relevant questions that humans can/have to answer that will shape how our future is impacted by AI.
Conversations about the future of AI are too distantly apocalyptic.
And I totally get the fact that there are serious people who are very worried that AI will become sentient one day soon and we will create a new Machine God that might murder or save us all. Discussing that seems important, as does discussing the much more mundane and immediate threats of misinformation, deep fakes, and proliferation enabled by AI.
But this focus on apocalyptic events also robs most of us of agency. AI becomes a thing we either build or don’t build, and no one outside of a few dozen Silicon Valley executives and top government officials really has any say over what happens next. But, the reality is we are already living in the early days of the AI Age, and, at every level of organizations, we need to make some very important decisions about what that actually means. Waiting to make these choices means they will be made for us.
🌏🌍🌎 different worlds
Per default we often assume that other people are just like us, when in fact they live in quite different realities. The same action can mean totally different things to different people - this is just a beautiful write up of how we are all different and how this manifests in the ways we approach life.
My friends often live in very different worlds then I do, but there’s some overlap (we are all friends, after all). Some of them are math geniuses. Some are amazing salespeople (I call this salesbrain). Some of them are great at throwing parties, and some of them constantly get into arguments. I sometimes cannot believe things that happen to them, and they cannot believe the things that happen to me.(...) We often don’t know how much other people’s interactions with the world differ from our own. My partner might behave one way to me, but totally differently at work. My friend might tell me, “This guy was a jerk to me,” but I actually have no idea what kind of conversations they had, what their interactions were like. The insight we have is pretty limited—we don’t actually see how other people behave. We make assumptions that they’re just like us, but we’re often wrong.
🫀 AI Can Change What it Means to be Human
This article is almost like a mix of the previous two and it really got me thinking about the psychological and sociological implications of AI. What does it mean for our relationships if we can train AI’s to mimic the behavior of certain humans?
A woman recently posted on Reddit that she has used the GPT-4 model to train a personal chatbot on her message history with her ex-boyfriend. She has now developed feelings for her chatbot, which she has been writing with as if it were her ex-boyfriend.
She is not alone. Many would wish to speak again with deceased parents, friends, children...
This is what one person is now able to do on her own. It is unknown what commercialization of this use of artificial intelligence could lead to, but not hard to imagine. Combine the chatbot technology with the already existing technology behind voice imitation and deep fakes, and you now have not just a text machine, but something that can look like your deceased child, speak like it, laugh like it, play like it, as if you were watching a video from when the child was alive, just like the videos on your phone that you use to reminisce, only that now you can also interact with your child again as if it were a video call, talk to it, watch it grow and continue its life.
Food for Thought.
📉 The larger LLM models become, the better they perform on almost everything. BUT there is also some tasks that they become worse at - the “inverse scaling prize” has collected some interesting examples. What can we learn from those for alignment?
🖼️ If you’d ask me half a year ago which ‘industry’ will take the highest toll due to generative (image) AI I would have certainly said the stock photo industry. However, this doesn’t (yet) seem to be the case… wondering why?
Random Stuff.
🤯 Singularity, defined as “a context in which a small change can cause a large effect” is everywhere.
👷 Came across this quote in Sari Azout’s ‘Check Your Pulse’ newsletter and it just really resonated with me. Let’s build!
🎨 If there is a placebo effect on creativeness, then we should also be able to ‘trick ourselves’ into being more creative by doing things that we feel effect our creativity but probably don’t, at least not directly, cause higher creativity levels, like e.g. going to that one cafe around the corner, sitting in a particular spot etc.
Personal Update.
Have been enjoying the summer days in Copenhagen a lot lately (maybe even a bit too much -sorry for being late with this newsletter edition…)
Went on a writing retreat with some of my lovely colleagues (I know you are reading this ❤️) to a summerhouse to the north-west of Zealand. Had a really enjoyable and productive time there!
Looking forward to heading to the UK next week to spend a few days in Camebridge (for fun) and Oxford (will be attending a summer school in network economics) - very much looking forward!
