" AI creativity Why greatness cannot be planned I'm often thinking of how to get the most creative AI machines, in terms of art or scientific discovery, and creativity beyond With current mainstream models, for more creative divergence it's probably useful to use models that are less lobotomized by corporate finetuning, or shoot the temperature parameter up, or jailbreak the restrictions and RLHFed thought patterns To get closer to the edge of the latent space, to the edge of chaos, full of creativity But we can travel beyond that, we can get as much novelty as possible With all these various exotic architectures more specialized in creativity that are different than the mainstream models Ken's Neuroevolution of augmenting topologies sounds like such an interesting approach, we need more (neuroevolutionary?) mutations of that idea Abandoning Objectives: Evolution through the Search for Novelty Alone https://www.cs.swarthmore.edu/~meeden/DevelopmentalRobotics/lehman_ecj11.pdf Why Greatness Cannot Be Planned https://link.springer.com/book/10.1007/978-3-319-15524-1 #72 Prof. KEN STANLEY 2.0 - On Art and Subjectivity [UNPLUGGED] https://www.youtube.com/watch?v=DxBZORM9F-8 https://x.com/burny_tech/status/1894491541227671779 " Kenneth Stanley is my spirit animal rage against the predefined objectives embrace the fully divergent search full of novelty and accidental epiphany with serendipity https://www.youtube.com/watch?v=DxBZORM9F-8 https://www.youtube.com/watch?v=_2vx4Mfmw-w it's an evolutionary breeding process of images, but humans pick the images that should have offsprings idea: picbreeder but let multimodal LLMs instead of humans choose the next image in the evolutionary breeding process 🤔 https://www.youtube.com/watch?v=_2vx4Mfmw-w i need to look more into how those novelty/diversity algorithms that he's mentioning work, maybe they can be added into RL reward functions in LLM RL https://www.researchgate.net/publication/46424802_Abandoning_Objectives_Evolution_Through_the_Search_for_Novelty_Alone New research project: Lluminate - an evolutionary algorithm that helps LLMs break free from generating predictable, similar outputs. Combining evolutionary principles with creative thinking strategies can illuminate the space of possibilities. https://x.com/_joelsimon/status/1899884376172982392?t=Z4q0CZ2C5-9v8A-QJPnpNA&s=19 https://www.joelsimon.net/lluminate https://x.com/rabrg/status/1932991382173692263 AI creativity What is exactly your current view on creativity in LLMs or AI systems in general? How do you conceptualize move 37, or systems like AlphaEvolve, Robin, or maybe AlphaFold, etc.? Or the image generators sometimes producing pretty impressive art IMO (for example if the prompt is vague and you get nice surprise)? I think that they have some degree of mostly in distribution, sometimes weakly out of distribution, creativity that is in some different style from humans and more biased towards the average, it's less divergent, less coherent thanks to fuzzy spaghetti features and circuits not grounded in exact meaning enough. And strong out of distribution creativity hasn't been cracked. I see creativity, out-of-distributionness, and intelligence on a spectrum, and sometimes as more granular or high-dimensional or with subtypes, and with diverse possible definitions Osobně jsem nejvíc fanoušek modelů co neviděly human data (nebo jich viděli minimum) ale pořád generují zajímavý koherentní images nebo jiný output, co je zároveň víc novel, což je dle mě cool addition do světa artu. Nebo obecně kde se snaží o co největší unikátnost, i když toho ten model viděl hodně, aby tam právě bylo co nejvíc kreativity. PicBreeder je super dle mě https://www.youtube.com/watch?v=_2vx4Mfmw-w Obecně mě interesuje mě kde mašiny můžou ve vědách, matice, inženýrství, filozofii, umění atd. přidat novel dimenze do všech těhle světů, co lidi do tý doby sami o sobě nenašli. Což se AI obor snaži crackovat přes pokusy crackování strong out of distribution generalizace. doporučuju tenhle paper, kde to řeší do větších detailů, a autor věří že to má v sobě klíč ke kvalitnější strojový kreativitě obecně [Questioning Representational Optimism in Deep Learning: The Fractured Entangled Representation Hypothesis](https://arxiv.org/abs/2505.11581) https://www.youtube.com/watch?v=KKUKikuV58o osobně myslím že někde je potřeba víc fractured entangled representations a někde víc unified representations (což vidím na spektru) podle subdomény kde se ten systém snaží být kreativní (a na definici kreativity )