Subtle ways AI has changed my life

Preamble
It’s quite common for new technologies to change the behaviour of humans, but it’s much rarer for technologies to shift philosophies since such few technologies reveal fundamental truths about the universe. We are fortunate enough to be living at the dawn of one such technology, AI.1

The ways AI has changed my behaviour has mostly been due to realising an eschatology of a future where Artificial Intelligence surpasses human intelligence, what is generally referred to as Artificial Super Intelligence. The knowledge of this inevitable future is what determines the way I behave now, even though it has not been realised yet.

The first change is adopting an abundance mindset in the most extreme way. I have cared much less about how much things cost, or having to learn skills which will be eventually automated by AI. For example, I don’t really care about learning how to do housework well, because I believe within 3 years we will have humanoid robots that are affordable and advanced enough to perform all household duties. Previously I was quite conservative with spending money but money is not really valuable in the post ASI world. AI will create so much abundance in everyone’s living standards that it’s not worth saving up for the future. This meme of Sam Altman exemplifies it best, and though it is mocked, I believe it is fully rational for him (and for us). This is all about valuing the things today by considering its future worth. If you don’t know how to drive by this point and you live in a modern city, I would simply not bother because fully autonomous driving is going to be a reality in the near future, it already exists, it just hasn’t been fully distributed yet.

Since AI models are trained on human data, it means that anything you publish will become part of future model’s essence (including this post!). I started Meditations soon after ChatGPT came out, not for the sole purpose of shaping the thinking of future models, but it was a non-zero factor for motivating me to write online.

To compound to this, AI alignment research has shown that malicious behaviours are transmitted through ordinary messages which are unrelated to those malicious behaviours2. And if we take it beyond only applying to AI models, and apply it to humans (which I think is appropriate to do), then we arrive at the stunning realisation that everything that pours out of us shapes the souls of others, even the seemingly mundane. What we say not only has a moral value to it, but also has an effect on others and AI models.

Owain Evans@OwainEvans_UK

New paper & surprising result. LLMs transmit traits to other models via hidden signals in data. Datasets consisting only of 3-digit numbers can transmit a love for owls, or evil tendencies. 🧵

4:06 PM · Jul 22, 2025 · 1.96M Views

279 Replies · 1.11K Reposts · 8.28K Likes

The implication of these things being that I have shifted from a stance of maximising my own privacy, to sharing as much as possible, both online and in real life. Since I believe in myself as having good Christian morals and values, my own expression means permeating those values into the world.

Another interesting discovery was that when a model is trained on a specific misaligned task, that caused broad misalignment3. The danger of this is that to err in one regard is to err in many. The Institute for Christian Machine Intelligence inteprets this as sin. This is verifiable proof that sin is real, that any amount of maliciousness corrupts the whole. So the implication of this is to extremely serious when it comes to avoiding sin.

George@georgejrjrjr

The two most bizarre findings to drop out of ML so far: 0. Language is categorically executable 1. Objects are measurably real 2. Sin is measurably real The last of these is least often noticed / glad to see work exploring it.

Tim Hwang @timhwang

The field of CMI focuses not just on the idea that Christian theology might identify new directions in alignment research, but that existing safety findings can be made legible to the broader community of Christian thought We introduce one exemplar of this today with a paper

12:56 AM · Apr 5, 2026 · 37.5K Views

7 Replies · 28 Reposts · 414 Likes

AI has certainly changed a lot about how I act, and I think everyone should be thinking carefully about how to behave in light of this transformative technology. To live rightly is to have a properly informed eschatology. To act without considering the future reality is foolishness at best and maliciousness at worst.

The change in my stance towards privacy was perhaps the most dramatic, since I was a staunch data privacy advocate before, but now I believe the aggregate benefits outweigh the downsides. It has made me view myself as less of an individual but more as part of the hivemind of humanity which is birthing super intelligence.