Chrisantha Fernando (@chrisantha_f) header
Chrisantha Fernando (@chrisantha_f) profile picture

Senior Research Scientist at Google #DeepMind, ex-Medical Doctor #WadhamOxford, PhD #Sussex, Artist.

Cromer, England
Joined February 2017
Photo by Chrisantha Fernando (@chrisantha_f)Photo by Chrisantha Fernando (@chrisantha_f)Photo by Chrisantha Fernando (@chrisantha_f)Photo by Chrisantha Fernando (@chrisantha_f)Photo by Chrisantha Fernando (@chrisantha_f)Photo by Chrisantha Fernando (@chrisantha_f)Photo by Chrisantha Fernando (@chrisantha_f)Photo by Chrisantha Fernando (@chrisantha_f)Photo by Chrisantha Fernando (@chrisantha_f)Photo by Chrisantha Fernando (@chrisantha_f)
Chrisantha Fernando retweeted
Why does the concept of 'red' not feel like experience of red? We argue this disconnect exists because our deep-seated need to be understood will always exceed the limits of the language we invent to achieve it. arxiv.org/pdf/2506.12086 #Consciousness #Neuroscience #Philosophy #AI
2
2
716
Chrisantha Fernando retweeted
[LG] Wanting to be Understood C Fernando, D Banarse, S Osindero [Google DeepMind] (2025) arxiv.org/abs/2504.06611
1
5
15
1,447
Chrisantha Fernando retweeted
“Wanting to be Understood” - Could a deep human need to be understood be the crucial evolutionary 'gadget' bootstrapping cooperation, culture and language? We explore this idea using AI simulations in our paper: arxiv.org/abs/2504.06611 🧠 #Evolution #Cognition #AI #GoogleDeepMind
3
8
3
22
3,774
Chrisantha Fernando retweeted
Amazing new work from @GoogleDeepMind ! The origin and function of external representations. — Dr. Chrisantha Fernando @chrisantha_f | Dr. Simon Osindero @sindero and Dr. Dylan Banarse @2ne1 journals.sagepub.com/doi/epu… External representations (ERs) are objects or performances in the world whose proper function is to communicate about other things in the world. Why and how did we make them, and what do they give us? We outline a simple framework for thinking about ERs grounded in modern machine learning. We propose a minimal set of neural mechanisms needed for open-ended ER production. Our constructivist enactive view contrasts with nativist views which propose specialist neural modules requiring symbolic internal representations. We propose a plausible set of (biological and cultural) evolutionary steps to full Gricean symbolic communication via a set of increasingly complex enactive algorithms for ER production. A pragmatic space of games is defined, which includes not only fully cooperative language games but also science, art, and evolved signal manipulation games. This space is defined by the complexity of learning needed by sender and receiver. We propose that one important step towards open-ended ER use was selection for bush reading, which like mind-reading is an inferential process requiring complex contextual and syntactic understanding of cues about events displaced in space and time. Bush reading pre-adapted receivers to be receptive, competent, and perspicacious interpreters of later intentionally produced signals about hidden topics such as felt mental states. This paved the way for minimally Gricean communication, which subsequently could be bootstrapped into explicit theories of mind, folk psychology narratives, and symbolic language in general. Recent findings in cognitive archaeology are integrated within the framework, and new experiments in machine learning suggested.
5
8
2,195
Cool video about our Promptbreeder: Self-Referential Self-Improvement Via Prompt Evolution (P... youtu.be/tkX0EfNl4Fc?si=Rl2-… via @YouTube #artificialintelegence #machinelearning #LLM
2
6
1,661
Chrisantha Fernando retweeted
Introducing Promptbreeder. Promptbreeder employs large language models like GPT-3 to iteratively improve text prompts. But here's the magic - it doesn't just evolve the prompts themselves. It also evolves how the prompts are generated in the first place. Let's break it down. Promptbreeder initializes a population of prompt variations for a task. It tests them out to see which perform best. The winners are "mutated" - modified in some way - and inserted back into the population. Rinse and repeat. But Promptbreeder makes the mutations smarter over time. It uses the AI to generate "mutation prompts" - instructions for how to mutate and improve a prompt. And it evolves better and better mutation prompts. So Promptbreeder is constantly getting better at getting better. It's a self-improving, self-referential loop, with natural language as the substrate. No messy neural network fine-tuning required. The results are prompts that are specialized and highly optimized for specific applications. On math, logic, and language tasks, Promptbreeder outperforms other state-of-the-art prompting techniques. This is a taste of the future. Soon AIs could help us find the right words, crystallize fuzzy ideas, and turn chaotic thoughts into elegant expressions. Promptbreeder demonstrates the potential for language models to be creative collaborators, not just passive tools.
31
261
35
1,513
475,084
Chrisantha Fernando retweeted
If you are blown away by "Take a deep breath" to be such an effective prompt for LLMs then check out Promptbreeder (xitter.gq/_rockt/status/17088293…). It found an even more effective and unintuitive prompt: "SOLUTION" works better and takes GMS8K zero-shot performance to 83.9!
In a new paper showing that AI comes up with more effective prompts for other AIs than humans do, there is this gem that shows how weird AIs are... The single most effective prompt was to start by telling the AI "Take a deep breath and work step-by-step!" arxiv.org/pdf/2309.03409.pdf
4
25
1
131
37,399
🌱 Introducing Promptbreeder: LLMs evolve their own prompts through self-referential self-improvement! Paper: arxiv.org/abs/2309.16797 #PromptEngineering #LLM #AI #ML #Promptbreeder
11
69
13
354
188,377
Promptbreeder evolves state-of-the-art prompt strategies on common benchmark maths, reasoning, and classification datasets. #PromptEngineering #LLM #AI #ML #Promptbreeder
1
11
2,874
Prompbreeder evolving a prompt to determine if a word (homonym) is used with the same meaning in two different sentences. #PromptEngineering #LLM #AI #ML #Promptbreeder
13
2,008
Chrisantha Fernando retweeted
You Only Live Once: Single-Life Reinforcement Learning abs: arxiv.org/abs/2210.08863
5
125
16
714
Exhibition of our art !!!!
A fab line up of @ArtAIFestival events for @BritishSciFest in #Leics 13-17 Sept, including @improbotics @eaedmonds Tomislav Stipancic @chrisantha_f & colleagues fm @DeepMind + @elluba on 13 Oct britishsciencefestival.org/ & more also at art-ai.io/programme
1
5
Chrisantha Fernando retweeted
Got to see @MirowskiPiotr presenting work w/ collaborators @2ne1 @MateuszOnAI @sindero @chrisantha_f -- CLIP-CLOP: CLIP-Guided Collage and Photomontage
2
2
10
Chrisantha Fernando retweeted
Tomorrow is the last day to submit to our panel "Visions of the future of human-machine creative symbiosis" at @RoyalAnthro conf between 6-10 June 🤖🎨 Organised with @MirowskiPiotr @chrisantha_f @korymath @arkitus More here: nomadit.co.uk/conference/rai… #creativeAI #AIart #AI
What’s interesting is that we have very little news coverage of the average Russians beliefs in Russia About what they think. One concern is the fact that nato has so much to lose. If it's the case that Russia has increasingly little to lose then one can ask will putin stop at u
1
1
kraine ? Would NATO really get into a nuclear war if Russia invaded a NATO country? I wonder. And putin probably wonders too.. if a hard line isn't taken., eg if the west isn't even willing to suffer a loss of oil and gas for this cause, what message does this send to putin about
1
1
the plausibility of mad ?! We need to suffer a bit now to prevent suffering later I'm afraid.
1
1
Chrisantha Fernando retweeted
Amrita Sher-Gil (1913-1941), right, influential Hungarian-Indian painter (here with her sister) #WomensArt
3
66
5
587