Mimicry not Meaning (Connection Problem S09E06)

Note: This is cross-posted from my newsletter in an attempt to both to make it easier to read this via RSS feed and to have this in my own independent archives. You can subscribe here to get it delivered to your inbox.


Hi there,

It’s been busy and a great start to the year, which is also the reason I haven’t sent any updates since the end of last year. I’m happy to be able to share that my work with Stiftung Mercator‘s Digital Society program will continue throughout 2023. I’ve been very much enjoying the work with Carla Hustedt and her excellent team. Over at the freshly renamed European AI & Society Fund, which I headed on an interim basis last year before the brilliant Catherine Miller took over permanently, we’ve also continued to work together. Finally, after a brief engagement last year, I’m excited to be working more intensely with Sovereign Tech Fund, a publicly funded program to strengthen public interest open source digital infrastructure ecosystems. I’m including a few links below.

Have a great start to the week,



Some quick links:


  • Agora Digital Transformation, the think tank we worked on for much of last year, is officially set up, but the team focuses on the backend / building out the organization for now (hiring, processes, etc.) for now, so my guess is that it’ll be a little while before seeing much public output. But you never know, so keep an eye out.
  • Last year, over at Stiftung Mercator, we commissioned a hands-on guide to more effective policy and advocacy work for civil society organizations. It outlined political processes at the German and European level, where to focus advocacy efforts to make a difference, and gave concrete advice on how to go about each step of the process. Why, you might ask? Increasing the capacity for policy work of especially younger, smaller digital civil society organizations is one of the key needs we’ve identified as key building blocks for a healthier digital civil society ecosystem. The German version is available on the Mercator website, the English translation is for now available here (it’ll be up on the website soon).

European AI & Society Fund:

Sovereign Tech Fund:
STF is growing & hiring. Currently, there are 3 open roles, with more to come soon:

With that said, on to the newsletter proper.


You’re receiving this because you signed up for this newsletter on tinyletter.com/pbihr or through my website, thewavingcat.com. The Waving Cat is my boutique research and strategic advisory practice focused on making sure that emerging technologies actually benefit society. I’m also on Twitter and Mastodon. If you’d like to work with me or bounce ideas, let’s have a chat.


Mimicry not meaning: Notes on the state of generative AI

Because ChatGPT et al still very much capture the zeitgeist. Over on The New Yorker, ChatGPT Is A Blurry JPEG Of The Web by Ted Chiang offers one of the most apt metaphors for thinking about generative AI I’ve come across.

In a similar vein, although approaching from a very different angle, is this piece New York Magazine has about chatbots and computational linguist Emily M. Bender. It’s ace. Like the Blurry JPEG metaphor, Bender lays out how there are inherent limits to what this type of generative AI can do, and failing to recognize these limits leads to all kinds of misunderstandings and general nonsense:

The models are built on statistics. They work by looking for patterns in huge troves of text and then using those patterns to guess what the next word in a string of words should be. They’re great at mimicry and bad at facts. Why? LLMs, like the octopus, have no access to real-world, embodied referents. This makes LLMs beguiling, amoral, and the Platonic ideal of the bullshitter, as philosopher Harry Frankfurt, author of On Bullshit, defined the term.

Bender also co-wrote (with Koller 2020) the Octopus Paper , which NY Magazine summarizes as follows:

Say that A and B, both fluent speakers of English, are independently stranded on two uninhabited islands. They soon discover that previous visitors to these islands have left behind telegraphs and that they can communicate with each other via an underwater cable. A and B start happily typing messages to each other.

Meanwhile, O, a hyperintelligent deep-sea octopus who is unable to visit or observe the two islands, discovers a way to tap into the underwater cable and listen in on A and B’s conversations. O knows nothing about English initially but is very good at detecting statistical patterns.

As you can imagine, mayhem ensues. Click on through to NY Magazine for details.

Related: Quick blog post of mine about exploring the boundaries of generative AI.

So, generative AI currently: Mimicry not meaning. Proceed with caution.


Charity as a Platform

Thanks to a pointer by Catherine Miller, director of the European AI & Society Fund, I revisited the 2018 exploration Charity as a Platform:

The idea of ‘Charity as a Platform’ is to create a common infrastructure for the social sector, based on the same principles as Government as a Platform — a platform of common tools, components and guidance.

The core idea, as best as I can summarize, is to make accessible more than funds, specifically to bundle “infrastructure”-type stuff to especially smaller non-profit organizations: Stuff like processing of donations, accounting, media relations work, etc.: This goes hand in hand with other thoughts that have very much gained in popularity among funders over the last few years that are sometimes bundled under the hard-to-spell “Funding+” (or Funding +; or Funding Plus), meaning anything you can do to help outside of money.

This is still very much a topic we are exploring now, across the board. Spending money is easy, but spending money well (in the context of philanthropy) is much harder. And creating offerings other than money is exponentially harder, because needs are so diverse. Some small orgs might need access to legal advice, others to media work, others to HR support. Some might just need a license for software or access to an expensive database. Other still might want or need some of these things but would rather not gain them through a funder, which is also totally legit. All this, of course, against the dual constraints of 1) not creating additional dependencies when the goal is to increase capacities and autonomy; and of 2) not DDoS-ing the funding organization or absorb too much funding that’s there to spend on supporting civil society by building up in-house capacity to offer these services.

Long story short, still very much working on this. Pointers very much welcome!


What will cities look like in a few years?

Super interesting bit (Guardian) about the “15 minute city” concept and how the urban planning theory has sent conspiracy theorists into a frenzy. Also, turns out young people aren’t big on cars.

Against the backdrop of Berlin’s mini drama around a few hundred meters of cars v cyclists v pedestrians on Friedrichstraße, and the overall culture wars around cars v cyclists in cities, just a few personal thoughts:

  1. Science says pretty clearly that there’s no future in (or with) fossil fuel powered cars, so this part of the debate seems extremely boring to me, like we just have to go through the motions, but the outcome is pretty clear. (The car industry, interestingly, appears to be on board already.)
  2. Within major cities, climate is only one aspect of why cars can be problematic. Inside cities, space has become contested, too. One car per person just doesn’t fly when public space is limited, so switching to EVs won’t change that at all. I personally am all on board for the Tokyo model in which there’s basically no public parking and if you cannot provide parking for your own car, you cannot own a car. Something has to give, and this approach seems more sensible than most. We need more space both for mobility for more people, which cars are in the way of; and more space to unseal the ground for climate resilience.
  3. Maybe the most important part: Life and mobility infrastructure is hugely different between cities and small town/countryside. In urban centers, public transport is (or can be) dense, distances short. In cities, the most promising solutions and approaches are relatively clear and known. In the countryside however, it’s nigh impossible to use those same approaches. Cars, or more customized (self-driving?) shuttles and many more creative solutions might be necessary. Approaches there are much less clear. So let’s fix what we already know how to fix, rather than wait until we know how to fix everything. I think it’s ok to move at different speeds. We can solve this for half the population now, and while we do we have time to figure it out for the other half of the population.

Anyway: Like in all things climate-related I personally believe my role to be pretty simple: To be the best possible ally to the younger people who fight for their future. “I don’t want to change my lifestyle” isn’t good enough if the damage is born primarily by other (younger) people.


If you’d like to work with me or have a chat to explore collaborations, let’s chat! 


Who writes here? Peter Bihr explores how emerging technologies can have a positive social impact. At the core of his work is the mission to align emerging technologies and citizen empowerment. He works as an independent advisor at the intersection of technology, governance, policy and social impact — primarily with foundations, non-profits and the public sector. Peter serves as special advisor to Stiftung Mercator’s Center for Digital Society, and as Interim Director for the European AI Fund. He co-founded ThingsCon, a not-for-profit that advocates for responsible practices in Internet of Things (IoT), and co-hosts the Getting Tech Right podcast. Peter was a Mozilla Fellow (2018-19) and an Edgeryders Fellow (2019). Based in Berlin, he tweets at @peterbihr and blogs at thewavingcat.com. Interested in working together? Let’s have a chat.


Know someone who might enjoy this newsletter? Please feel free to forward your copy or send folks to tinyletter.com/pbihr.

Leave a Reply