Peter Gasston's Avatar

Peter Gasston

@petergasston.co.uk

Londoner, digital technology enthusiast, generalist, humanist, protopist, left-of-centrist, not a futurist.

133
Followers
61
Following
555
Posts
05.07.2023
Joined
Posts Following

Latest posts by Peter Gasston @petergasston.co.uk

It’s a source of constant amusement to me that there are two open, federated microblog protocols that are completely incompatible with each other.

09.03.2026 23:18 πŸ‘ 0 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0

I'm in a supermarket in Czechia buying some beers, and I'm chatting with Gemini Live + camera input to identify the different types and tell me all about them, and the objectors who refuse these tools are missing out on tiny miracles like this.

05.03.2026 17:41 πŸ‘ 0 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0
Preview
The Future of AIΒ Images Advancements in AI image generation, particularly through Google's Nano Banana Pro, highlight the need for models to understand causality and physics. Current models struggle with sequences, failing to accurately depict actions like a falling wine glass. Future breakthroughs may involve world models that comprehend physical laws, enhancing image realism.

Google’s Nano Banana Pro was a breakthrough in image models because of its world knowledge. What will be the next breakthrough? I think it has to do with the β€œfalling wine glass” problem…

02.03.2026 11:51 πŸ‘ 0 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0
Preview
Nano Banana 2: Combining Pro capabilities with lightning-fast speed Our latest image generation model offers advanced world knowledge, production-ready specs, subject consistency and more, all at Flash speed.

Google just launched #NanoBanana 2, a faster version of its image generation / editing model. N🍌2 replaces N🍌 Pro in the Gemini app, although paid users will be able to regenerate images using Pro. #AI
blog.google/innovation-a...

26.02.2026 16:34 πŸ‘ 1 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0

In the absence of strong evidence, teen social media bans seem to be an example of the politician’s syllogism: something must be done; this is something; therefore we must do this.

25.02.2026 14:30 πŸ‘ 3 πŸ” 0 πŸ’¬ 1 πŸ“Œ 0
Close-up photo of a Jackdaw looking directly at the camera

Close-up photo of a Jackdaw looking directly at the camera

Jackdaw (Corvus monedula) β€” a lucky pic at just the right moment. #birds

24.02.2026 16:17 πŸ‘ 0 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0

this essay has been going around and lemme just say it's an amazing piece of hard science fiction -- like, it almost redefines the genre into something that includes financial projection -- but that's all it is

we are in an era when AI panic porn is widespread and less than helpful

23.02.2026 23:52 πŸ‘ 84 πŸ” 10 πŸ’¬ 6 πŸ“Œ 1
Preview
'Image manipulation has always been around': 10 early photographic 'fakes' that trick the eye A century and a half before today's AI deep fakes, photographers created remarkable image manipulations. Here are 10 images from the 19th and 20th Centuries that tricked the viewer.

… which I found via this very interesting BBC piece on the history of fakes in #photography.
www.bbc.co.uk/culture/arti...

24.02.2026 10:01 πŸ‘ 2 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0
Preview
Composite Imagery and the Origins of Photomontage, Part I: The Naturalistic Strain IN 1931, RAOUL HAUSMANN, one of the original members of the Berlin Dada Movement, gave a public lecture on the nature and history of photomontage,^^1^^ on the occasion of the…

That’s from The Photographic News, published in 1861 but still strangely relevant today ( #AI ). Quoted in the article Composite Imagery and the Origins of Photomontage, Part I, in ArtForum…
www.artforum.com/features/com...

24.02.2026 10:01 πŸ‘ 0 πŸ” 0 πŸ’¬ 1 πŸ“Œ 0

β€œA photographer, like all artists, is at liberty to employ what means he thinks necessary to carry out his ideas… but let it be clearly understood, these are only means to the end, and that the picture when finished must stand or fall entirely by the effects produced, and not by the means employed.”

24.02.2026 10:01 πŸ‘ 0 πŸ” 0 πŸ’¬ 1 πŸ“Œ 0
Preview
SeeDance and the new mediaΒ landscape A few months ago we were pondering a few things about the future of media here at Llama Towers. I speculated about a world where you could generate your own Star Wars trilogy based on the Timothy Zahn's novels, fix the ending of Game of Thrones, or finish cancelled series like Firefly. I argued that copyright was increasingly ill-suited to this future, and that image and personality rights would become the real legal battleground.

SeeDance and the new mediaΒ landscape

A few months ago we were pondering a few things about the future of media here at Llama Towers. I speculated about a world where you could generate your own Star Wars trilogy based on the Timothy Zahn's novels, fix the ending of Game of Thrones, or finish…

19.02.2026 06:04 πŸ‘ 7 πŸ” 3 πŸ’¬ 1 πŸ“Œ 0
Post image

I could listen to a track or two out of morbid curiosity but a whole album would be too much.

19.02.2026 16:50 πŸ‘ 0 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0

A new version of the play β€˜Cyrano de Bergerac’, but the character of Cyrano is replaced by #ChatGPT.

18.02.2026 11:21 πŸ‘ 0 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0
Preview
A Guide to Which AI to Use in the Agentic Era It's not just chatbots anymore

Every few months, I write an updated, idiosyncratic guide on which AIs to use right now.

My new version has the most changes ever, since AI is no longer just about chatbots. To use AI you need to understand how to think about models, apps, and harnesses. open.substack.com/pub/oneusefu...

18.02.2026 01:50 πŸ‘ 129 πŸ” 30 πŸ’¬ 5 πŸ“Œ 8
Video thumbnail

The new ByteDance SeeDance 2.0 video model is VERY good. Each video is the very first output of the prompt. There are four, worth seeing them all to get a sense of the range (and potential issues)

"A nature documentary about an otter flying an airplane"

11.02.2026 03:07 πŸ‘ 145 πŸ” 19 πŸ’¬ 17 πŸ“Œ 17

I should add that this stuff was basically science fiction five years ago so even getting something this good should be viewed as a minor miracle.

10.02.2026 14:58 πŸ‘ 0 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0
Video thumbnail

#Kling3 has excellent prompt adherence, visual detail, and consistency across shots, but its motion and physics are off. There’s something wrong in every generation I’ve made so far, and it’s too slow and expensive to just run multiples until I get a good one. #AI

10.02.2026 14:56 πŸ‘ 0 πŸ” 0 πŸ’¬ 1 πŸ“Œ 0
A demonstration of where you can find the drafts feature in the top right corner of the post composer on Bluesky.

A demonstration of where you can find the drafts feature in the top right corner of the post composer on Bluesky.

v1.116 is rolling out now!

For all the overthinkers and perfectionists out there,Β we're launching Drafts.

09.02.2026 20:21 πŸ‘ 15847 πŸ” 3435 πŸ’¬ 920 πŸ“Œ 2470

Genuinely can't believe how much he's blaming missing players, considering the size of their squad. FFS

03.02.2026 22:28 πŸ‘ 1 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0

The commentators thrive on upsets, I get that. But tonight was beyond the pale.

03.02.2026 22:26 πŸ‘ 0 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0

Did he really say that? Because that's not what he said at half time, the fucking prick.

03.02.2026 22:08 πŸ‘ 1 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0

Gary Neville! Jamie Redknapp! Wonder kid EstevΓ£o! Ice cold Cole Palmer! Liam Rosenior’s tactical masterclass! Your boys took a hell of a beating!

03.02.2026 22:05 πŸ‘ 0 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0

UP YOUR HOLE GARY NEVILLE

03.02.2026 22:00 πŸ‘ 3 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0

I just had to switch to ITV, couldn't stand it anymore.

03.02.2026 21:19 πŸ‘ 1 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0
The researchers evaluated several major large language models, including ChatGPT, Claude, Gemini, and others, and compared their results with data from 100,000 human participants. The outcome marks a clear turning point. Some AI systems, including GPT-4, scored higher than the average human on tasks designed to measure divergent linguistic creativity.

β€œOur study shows that some AI systems based on large language models can now outperform average human creativity on well-defined tasks,” explains Professor Karim Jerbi. β€œThis result may be surprising β€” even unsettling β€” but our study also highlights an equally important observation: even the best AI systems still fall short of the levels reached by the most creative humans.”

The researchers evaluated several major large language models, including ChatGPT, Claude, Gemini, and others, and compared their results with data from 100,000 human participants. The outcome marks a clear turning point. Some AI systems, including GPT-4, scored higher than the average human on tasks designed to measure divergent linguistic creativity. β€œOur study shows that some AI systems based on large language models can now outperform average human creativity on well-defined tasks,” explains Professor Karim Jerbi. β€œThis result may be surprising β€” even unsettling β€” but our study also highlights an equally important observation: even the best AI systems still fall short of the levels reached by the most creative humans.”

A new study finds that #AI can outperform average human creativity in test conditions β€” but not above-average human creativity.
scitechdaily.com/ai-is-now-mo...

02.02.2026 17:04 πŸ‘ 0 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0
Post image Post image

ChatGPT is still the most-used #AI platform, but it's plateauing / in decline. Gemini is growing.
(Data is based on desktop Web use; apps may tell a different story).

02.02.2026 16:59 πŸ‘ 0 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0
Post image

Name the film.

29.01.2026 17:31 πŸ‘ 0 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0

think this might be one of the first good instances of AI visual art I've seen. the making of is great too cannoneyed.com/projects/iso...

29.01.2026 16:11 πŸ‘ 20 πŸ” 4 πŸ’¬ 0 πŸ“Œ 0
Preview
London is far safer than violent viral videos will have you believe Contrary to social-media lore, it is one of the safest cities in the world

Contrary to social-media lore, it is one of the safest cities in the world

28.01.2026 16:05 πŸ‘ 74 πŸ” 26 πŸ’¬ 2 πŸ“Œ 3

In the unlikely event anyone wants to read some words on a topic adjacent to the below, here are some (by me, sorry) www.thenewworld.co.uk/matt-muir-do...

28.01.2026 12:51 πŸ‘ 14 πŸ” 5 πŸ’¬ 1 πŸ“Œ 1