photog.social is one of the many independent Mastodon servers you can use to participate in the fediverse.
A place for your photos and banter. Photog first is our motto Please refer to the site rules before posting.

Administered by:

Server stats:

246
active users

#cybernetics

0 posts0 participants0 posts today

Here's a lovely piece by my #ANU #Cybernetics colleague, @theEllamo which talks about my #TokenWars talk, and how it's related to concepts like #PeakToken and the value of human-generated #data as the internet becomes polluted by #AI-generated slop.

There's a video link here to the #TokenWars talk, if you haven't seen it already.

Thanks, Ella!

cybernetics.anu.edu.au/news/20

ANU School of Cybernetics · Token WarsPhD Researcher Kathy Reid (she/her) is an AI Voice researcher investigating speech technologies with a focus on the data that goes into these models. Kathy asks critical questions about these technologies, the people, and the voices they serve. Kathy’s motivation both personally and professionally is the value that knowledge is power and knowledge shared is empowerment. These underlying understandings are core to Kathy’s PhD work and also her keynote talk Token Wars. With these values and a highly open-source background it may come as a surprise to you that Kathy questions if everything should be open not just to anyone but to anything. The continuous scraping and pollution of the internet by AI companies looking to train their latest models is deeply challenging to the ‘everything open’ approach. Kathy and her research ask a lot of great questions about technology and power - who benefits from technologies and what are the costs? These cybernetics questions about unintended consequences underpin Kathy’s Token Wars, a talk that dives into the current technical, legal, and political, resource conflict surrounding AI training ‘tokens’ or data. Kathy’s talk, as many great talks do, comes in three parts: Part 1: Kathy gives us an accessible overview of tokens and transformers, the technologies that together build large language models like ChatGPT and Claude Part 2: Kathy unpacks the value of tokens, why they mean so much to AI companies, and what it means for these tokens to become a scarce resource. Here, Kathy also dives into the actions and intentions of the key actors in these token wars, as well as the damage they are causing. Part 3: Kathy considers tokens and data as a form of treasure or capital – and asks how we might protect and safeguard this treasure. Kathy also speculates on the future of tokens and future protection strategies. The Token Wars: why not all our content should be open Token Wars was first delivered by Kathy Reid at Everything Open and the Melbourne Machine Learning and AI meetup. This version of Token Wars was delivered on Ngunnawal and Ngambri Country here at the Australian National University’s School of Cybernetics. Current LLMs are trained on nearly all the publicly available data in the world – and globally we’re running out of new human-generated ‘tokens’ to train newer and better models on. Kathy holds that we’ve passed a point in history she terms “Peak Token” - where we have the highest availability of human-generated tokens. As LLMs and synthetics data proliferate, the open web is becoming increasingly filled with low-quality “AI slop”, ushering in the “slopocene” - where rich, diverse, human-generated data is rarer and more valuable. LLM Models: Number of training tokens and parameter size by date. 2025-04-13 https://github.com/KathyReid/token-wars-dataviz. Visit Kathy’s blog for her recent thoughts on the speculative OpenAI hardware device that may come about as a result of these Token Wars and find out more about Kathy’s research on our PhD Spotlight from earlier this year.
Continued thread

Some quotes from the text:

"The goal is always to optimize the fulfillment of needs while keeping decisions and data as local as possible."

"Or in complex system lingo: The constraints are slightly different, which might result in the systems sliding into different attractor states."

"We can make use of the ideas of cybernetics if and when they seem useful to solve a problem, but they don't force us to act or think in a given way."

I wrote down some reflections on the complex anarchism symposium I attended in May and the talk I gave there about anarcho-communist economics. The text covers discussions around the question of which concepts and tools should be discarded as too central or too dominant in the anarchist context and which structures and tools are required to adequately handle complex systems in a non-hierarchical way.

transform-social.org/en/texts/

transform-social.orgTransform Social | Reflections on Complexity and CentralityThis is a reflection on the experiences from and feedback received at the complex anarchism symposium. It covers discussions around the question of which concepts and tools should be discarded as too central or too dominant in the anarchist context and which structures and tools are required to adequately handle complex systems in a non-hierarchical way.
Replied in thread

@Di4na @bert_hubert

Ahh, roger, thank you! That's a really useful explanation; yes, I see why slapOS in and of itself won't cut the mustard. Gosh, I can see why we need this capability from a strategic/geopolitical point of view. Indeed not just Europe; any competent government should want to make sure feedback-loops were set up so that they did not inadvertently lose "digital sovereignty" in such a manner here illustrated.

I did this about a decade and a half ago

It is mostly about defining the self; the illusion of control; what luck is; what affects our thinking; the attention economy; the literacy virus; the motor that motivates us

It’s a recap of everything I’d been writing about and teaching, up to that point – shortly after this I did a few actual personal appearances in venues to deliver this talk to audiences.

Success In Seven Parts – 18 mins 14 seconds, © Ian K Tindale 2011
youtu.be/6V6jKRF9B4s #Self #Control #Cybernetics #Luck #Literacy #LiteracyVirus #Attention #AttentionEconomy #Motivation

youtu.be- YouTubeEnjoy the videos and music you love, upload original content, and share it all with friends, family, and the world on YouTube.
Replied in thread

@xris @HadasWeiss

Couldnt get a clear grip on your interests from your feed, so came here to say this too. Also markets ( #invisibleHand - theory of moral sentiments, and the wealth of nations) , #qualityControl, and #openBSD, I reckon. I'm delving these at the moment and it's super interesting. Also Enterprise Resource Planning software suites (specifically #ERP5 but it might prove too deep for my delving ability) and sometimes #cybernetics (which goes back to #mathematical #complexity and #communicationTheory ).

academia.edu/31793918/Walter_A

adamsmithworks.org/texts

man.openbsd.org

www.academia.eduWalter Andrew Shewart, Economic Control of Quality of Manufactured ProductWalter Andrew Shewart, Economic Control of Quality of Manufactured Product
Replied in thread

@tante
…and they are trying to bring back #nuclear with #AI

While #nuclearenergy is not #fossil it not #green at all - but #colonial , #deadly & totally unfair because, like in AI, the #profits are being privatised, the costs are hitting the whole #society & the #planet

But from the point of view of someone believing in #cybernetics or one of its #TESCREAL grandchildren it is great - because in cybernetics EVERY problem will be solved in "the future" by "technology"

Replied in thread

@jalefkowit

It's an interesting point you highlight because I've never thought of things through that paradigm. Now that I see it, I realise I'm hoping to transition to #openBSD because I believe of mainstream operating systems, it probably has the best #DX.

In this particular case, I believe that the developers exploit that in order to optimise for what might be called #security which impresses me.

I realise it's probably an outlier in this regard!

I realized I have very little idea of the early reception (and development ?) of cybernetics in The Netherlands, but this is the earliest text I found so far: Fred Polak was the director of the Central Bureau of Economic Planning and in the late 1940s he travelled to the US where he was very much impressed by the first computers and the work of Wiener, Rashevsky and Shannon.