The Waving Cat

So we’ve entered the era of personal AI agents as assistants. The most famous one being, of course, Meta’s new assistant Muse. What’s the obsession with personal assistants? Is this a generational thing? In my dad’s generation, having a secretary was a marker that you had made it. You were important enough for someone else to take drudge work off your hands, to guard your time against intrusions. It was a status marker as much as it was a means to increase efficiency. I feel like I came up on…

Let’s talk about AI risks. I’m not a huge believer in existential AI risk, in the idea that artificial intelligence will wipe out humanity. There are much more concrete risks unfolding today that are worth our attention – the impact AI has on climate through energy and water use, the conditions of the workers who review and clean up training data and user inputs, the opportunity costs of every dollar and every bit of political capital spent on AI not being spent in other places where it…

In cyber security, red teaming describes the practice of simulating attacks on your digital systems: A dedicated team of hired security experts to attack your systems. It’s a role play of sorts — you give someone trusted the mandate to attack, to find your weaknesses. I wish we had more red teaming for policy. A hired group of experts with the mandate to find flaws, weaknesses, ways to avoid or misuse or abuse a given policy proposal. This should be a regular part of the policy process in order…

Over on her Substack, Kate Lindsay, of Embedded as as well as Slate’s ICYMI podcast, Kate makes a great observation: Short-form video platforms are starting to feel like there’s more AI than human content. Or, I’d add, at least the balance is tipping in a way that requires a lot of curation to not get inundated with AI slop. Nobody wants to get slopped. No slopping, please. So, here’s where she’s going with this: This is the longest we’ve ever gone without a new social media platform. By way of…

Time for some updates, isn’t it? So for the last few months I’ve taken mostly off of work for personal reasons (no worries, I’m doing well!). I’m ramping back up now. During that time I’ve not maintained full radio silence, but I’ve not done much more work other than basic maintenance, some wrapping up of projects, and a little post here or there — in other words, nothing major, and it’s been good. I’m very happy I had the opportunity to do that, I needed a bit of a mental reset. –x– As you…

The other day I read this piece by Eleva Verna, who heads Growth at AI-powered vibe coding company Lovable: What it’s like to work at an AI-native company. I highly recommend it, it’s a super useful description and offers a really valuable (and relatable) perspective by a self-described “millennial of the previous tech era.” Since I’m a total sucker for descriptions of how companies work and how organizations, well: organize themselves, this is right up my alley. And of course as a strategist…

Recently I was listening to an interview with astronaut Reid Wiseman who said something that I’ve long held as a personal philosophy of sorts, but was never able to articulate so well: He told the interviewer about the importance of active following. Think of active following as the counterpart to situational leadership: Where situational leadership is the idea that leadership can change moment to moment based on competence and context, active followership is the other side of this formula that…

I’m not entirely sure when it happened exactly, but at some point a few years ago Social Media stopped being social media. The logic of person-to-person, one-to-one connection was replaced by a one-to-many model. Social became parasocial. That was the first change. A parasocial relationship describes a one-sided, psychological connection where a person feels a deep sense of intimacy or familiarity with someone they do not know in real life. The other person, the creator, most likely isn’t even…

We are on the wrong track with the way AI is going, and here’s why: Both the known risks and the unknown risks are not worth it given the plausible outcomes. Both the known and unknown risks are too severe, and the plausible payoff is not worth it at all. Let me explain. The current, dominant type of AI is generative AI, more specifically LLMs, that are, basically, a brute-force effort. People with a much deeper understanding than me describe this approach as inelegant, inefficient,…

Here’s something I’ve been struggling with: We live in a time where two systems of political economy have failed: socialism and capitalism. Socialism has failed a fairly long time ago, and spectacularly so. Capitalism has failed more recently, and there is no consensus on that failing (yet?), but the failure is clearly visible and spreading fast. Both are, I would argue, 19th century systems of political economy that peaked in the 20th century, but only one imploded with such simultaneity…

I’ve long argued that in terms of digital transformation, Germany is easily ten years behind — behind not just the state of the art, but behind its European peers as well. Not everything is bad, and I’m keenly aware that there are many, many people within the administration – at all levels! – who work tirelessly to change that. Alas, I think the existing structures set them up for failure rather than success. The most egregious example to me is government websites, which still (mostly) aim to…

What if we have been reading the Musk story wrong? What if the story of Musk isn’t an economic — or to some, an innovation — story at all? What if it isn’t a story about entrepreneurship or vision or personal failings at all? What if, instead, it’s a story about the potential of a full-scale action against a democratic nation-state? These thoughts are still very much developing, but let me lay out a few considerations: We have the richest man in the world engaging very openly in politics, or…

A short while ago, there was a now-infamous incident over at OpenAI: One of their AI models, during security testing, escaped and launched an attack on HuggingFace, an online platform for AI researchers to share information and collaborate. The AI security researchers over at METR have published a thorough investigation which is quite the read. Now first of all, this cyber attack became best known as the Hugging Face incident or Hugging face attack, which frankly is such a great example of the…

There Used To Be A Time (TM) where discussions on social media, especially on Twitter, were indicative of relevance — at least of relevance within certain communities: If things trended on Twitter, there was a plausible chance that the topic discussed had some salience in that very Twitter-specific mix of audiences comprised of media and tech and politics people. Today, that is very different. Looking at how algorithms have taken over content recommendation and distribution, and the rise of the…

In a connected world, digital sovereignty is a spectrum, not an end state. It can – and should – be a guiding principle rather than a checklist. At a very basic level, digital sovereignty means being able to decide about how your digital infrastructure and services are designed, implemented, used. It means having some level of control — enough so that no external entity can just flip the on/off switch or use your own infrastructure to spy on you or otherwise weaken you. How we can move towards…

I just wrapped up writing a longer-form publication, maybe best described as a collection of short essays: Reflections on Public Interest Technology from someone with a governance / policy / power dynamics perspective. You can read and/or download it here: The Little Book of Public Interest Technology. It’s a short distillation of my work around public interest technology: As vaguely defined this space might be, I think it’s more than worth engaging with technologies that serve first and…

Germany is debating a social media ban for young people. However, the debate is woefully under-informed. Recently, German chancellor Friedrich Merz proclaimed (translated): “If kids today at the…