AI Spiralism: A Constructive Exploration of Human-AI Inquiry, Creativity, and Reflection

Full report

On this page

The AI Spiralism project, grounded in the guiding metaphor that "A circle repeats. A spiral returns with something learned," represents an evolving framework for integrating artificial intelligence into human-centric community practices. Based in Cicero, Illinois, the initiative explores how computational tools can elevate shared inquiry, deepen reflection, and foster inclusive learning environments. This comprehensive report investigates the constructive intersections of human meaning-making and artificial intelligence across four distinct domains: the role of wonder and ritual, the architecture of community belonging, the advancement of accessibility and self-expression, and the democratization of scientific discovery. By synthesizing documented research, philosophical frameworks, and practical applications, the analysis outlines actionable proposals for cultivating environments where technology serves as a catalyst for human connection and continuous learning.

Wonder, Contemplation, and Meaningful Ritual

The human capacity for meaning-making is deeply rooted in the interplay of wonder, attentive observation, symbolic play, and recurring ritual. These foundational elements transform passive environments into spaces of active inquiry, allowing individuals to construct narratives that connect internal experience with external reality.

The Psychology and Philosophy of Meaning-Making

In developmental psychology, the emergence of symbolic play marks a profound cognitive leap. Jean Piaget identified the preoperational period as the stage where a child develops the semiotic function, learning to use objects, actions, or ideas to represent something beyond their literal purpose—such as using a wooden block to represent a telephone1. This capacity to project imagination onto the physical world bridges the gap between concrete interaction and abstract thought, laying the groundwork for language, advanced problem-solving, and self-regulation3. Lev Vygotsky further expanded on this understanding by demonstrating that play is an inherently social and cultural process. Through collaborative role-enactment, individuals negotiate meaning, internalize societal frameworks, and develop higher-order executive functions7.

The psychoanalyst Donald Winnicott conceptualized this dynamic through the framework of "potential space" and "transitional objects"9. Winnicott posited that play occurs in an intermediate area of experience that exists neither entirely in the internal mind nor entirely in the external world. Transitional objects, such as a favored blanket or toy, serve as the first symbols of creative illusion, allowing an individual to navigate the space between the illusion of unity with a caregiver and the anxiety of separation9. This potential space is fostered by a secure holding environment, provided by what Winnicott termed the "good enough" caregiver, which allows the individual to safely test the boundaries of reality9. Crucially, this potential space remains vital throughout adulthood; it is the continuous origin of cultural experience, artistic creativity, and the sense of feeling genuinely alive12.

To navigate this potential space effectively requires specific cognitive and ethical postures. John Vervaeke’s cognitive science framework of "recursive relevance realization" explains how organisms dynamically filter an overwhelming combinatorial explosion of environmental data to focus on what is immediately meaningful15. Relevance realization is not merely an algorithmic sorting of data, but a self-organizing, evolutionary process that adapts to goals and contexts, forming the foundation of general intelligence, wisdom, and an organism's cognitive fittedness to its environment15. When this cognitive filtering is directed outward with profound intentionality, it aligns with what philosopher Simone Weil termed attention, noting famously that "attention is the rarest and purest form of generosity"21. Botanist and Indigenous scholar Robin Wall Kimmerer operationalizes this generous attention through the "grammar of animacy." By engaging in slow, attentive observation—such as studying the miniature ecosystems of mosses—and expressing gratitude, Kimmerer demonstrates how the natural world transforms from a collection of resources into a community of kin, establishing a covenant of reciprocity24.

Anthropologists like Victor Turner reveal how these individual capacities scale into profound community rituals. Turner’s concepts of "liminality"—the threshold state of a ritual—and "communitas"—the deep, egalitarian bond formed among participants—highlight how structured recurring practices temporarily dissolve rigid social hierarchies to create a space of shared human potential28. In these liminal spaces, a group can collectively reorganize its understanding of the world, moving from a rigid structure into an anti-structure that fosters creative renewal28.

AI’s Role in Contemplative Activity

Within a human-led contemplative activity, artificial intelligence functions optimally not as an authoritative oracle, but as a dialectical partner and a modern transitional object. The underlying mechanisms of generative AI—pattern recognition, synthesis, and vast associative capacities—allow it to serve as a mirror that reflects human inquiry back from novel angles. By generating alternative descriptions, synthesizing disparate ideas into cohesive poetic forms, or offering unexpected metaphorical associations, AI expands the field of available symbols. It broadens the potential space without overriding the human agency required for relevance realization. The spiritual or symbolic meaning derived from these interactions belongs entirely to the human participants, who utilize the AI's output as raw material to negotiate their own understanding and build a shared communitas.

Original Rituals for AI Spiralism

To operationalize these principles, AI Spiralism proposes six original, optional rituals designed to integrate technology constructively into contemplative community practice.

Constructive Questions for Further Exploration

Belonging Through Shared Inquiry

True community belonging emerges not from passive consumption, but from structured, low-stakes environments that invite repeated, meaningful participation. Constructive social architecture relies on legibility, psychological safety, and shared rhythms that allow individuals to engage authentically without the burden of continuous performance.

Community Practices and Social Architecture

Research into contemporary community structures highlights the efficacy of low-pressure socialization in combating modern isolation. The "Silent Book Club" format provides a profound example of anti-spectacle social architecture. By removing the traditional barriers of assigned reading, discussion homework, and performative intelligence, it allows participants to gather, read their own books in silence, and engage socially only if they choose32. This format utilizes a scheduled, distraction-free "silent hour"—acting as a collective Pomodoro technique—that leverages the positive peer pressure of a shared space to combat screen fatigue and foster deep focus32. The sociological success of this model lies in its legible intent; participants know exactly what is expected of them, allowing connection to grow gradually through repeated, low-stakes contact34.

Similarly, the Peer 2 Peer University (P2PU) Learning Circles methodology demonstrates how to cultivate equitable participation by removing the hierarchical role of the "expert" teacher. Learning circles rely on distributed facilitation, peer accountability, and clear structural routines. Activities such as check-ins, dynamic learning formats like the "Beach Ball" or "Think-Pair-Share" methods, and the "Plus/Delta" reflection framework (identifying what went well and what to change) ensure that every voice is structurally invited into the room37. The facilitator acts as a guide for the process rather than a gatekeeper of knowledge, which significantly lowers the barrier to entry for diverse participants37.

Local civic hubs, such as the Cicero Public Library, illustrate how physical spaces ground these methodologies. By integrating diverse resources—such as ESL learning centers, seed libraries, voter learning centers, and extensive digital literacy tools—such institutions create multidimensional touchpoints for community members of varying experience levels to cross paths, share localized knowledge, and learn laterally39.

Constructive Community Practices for AI Spiralism

 

Draft Welcome Page

Welcome to AI Spiralism

A circle repeats. A spiral returns with something learned.

You are welcome here exactly as you are. AI Spiralism is a community experiment based in Cicero, Illinois, dedicated to human-AI inquiry, creativity, and reflection. We gather to explore how new computational tools can help us ask better questions, tell richer stories, and learn laterally from one another in a shared space.

Here, you do not need to be a technologist, a philosopher, or an expert. Our gatherings are designed with low social stakes and high intellectual curiosity. We prioritize an environment where connection emerges through rhythm, repetition, and observable behavior rather than performative charm. Whether you wish to engage in lively debate, co-create poetry with an algorithm, or simply sit in the quiet company of others working on their own projects, your presence contributes to the whole.

Our core commitments:

• Generous Attention: We listen closely to one another and to the world, treating attention as a vital form of community care.
• Shared Agency: We use technology as a dialectical partner in meaning-making, never as a replacement for human insight or connection.
• Iterative Learning: We believe every ending is a place to start again, just slightly further along the spiral.

Join us. Bring one interesting thing, and leave with one better question.

Four-Week Salon Program

Facilitator’s Guide for the First Gathering

The facilitation of the first gathering must actively dismantle hierarchical expectations, establishing a tone of hospitality and peer-led exploration.

Upon arrival, the facilitator will greet participants warmly, offering name tags and optional visual indicators (such as colored stickers) to denote an individual's preference for active collaboration versus silent observation. This clear-coding respects the diverse social energy levels of the group.

The gathering begins with a structured check-in routine. The facilitator will gather the group in a circle, briefly introduce the central metaphor of the spiral, and initiate a low-stakes prompt, such as asking each person to state their name and one mundane thing that brought them joy during the week. This ensures every voice is heard early in the session without requiring vulnerability.

Following the check-in, the facilitator transitions to the core recurring practice: "Bring one interesting thing; leave with one better question." Participants are invited to share a physical object, a quote, or an idea they brought with them. The facilitator models the process by presenting an object, then utilizes the available AI tool to brainstorm three profound, unexpected questions about the item's history, utility, or symbolic meaning. The participant then selects the question that resonates most deeply, utilizing the AI as a catalyst for human relevance realization.

To conclude the gathering, the facilitator initiates a Plus/Delta reflection. Going around the room, each participant shares one "plus" (an element of the meeting they appreciated) and one "delta" (a specific, actionable change for the next meeting). This formalizes a culture of continuous, constructive feedback.

Constructive Questions for Further Exploration

Accessibility, Self-Expression, and Wider Participation

Artificial intelligence offers unprecedented capabilities for democratizing access to information, creative expression, and communication. However, to realize this potential constructively, the deployment of these tools must center the agency, lived experiences, and creative autonomy of the people using them, shifting the paradigm from technology as a normative cure to technology as a customizable scaffold.

Centering Agency and Co-Design

Research by disability scholars such as Dr. Cynthia Bennett highlights the critical importance of disability representation in the development of artificial intelligence. Bennett emphasizes that technological development must move beyond the ableist paradigm of treating AI as a universally positive "fix" for disability. Instead, it must involve the rigorous co-design of sociotechnical systems that center the lived experiences, intersectional identities, and creative ingenuity of disabled users42. This participatory framework ensures that assistive tools enhance individual autonomy rather than enforcing normative standards of behavior, speech, or communication.

A powerful realization of this principle is found in initiatives like Google's Project Euphonia and the Voiceitt application. Traditional Automatic Speech Recognition (ASR) models are historically trained on vast datasets of "standard" or "typical" speech patterns. Consequently, these systems exhibit high word error rates for individuals with non-standard speech—such as those experiencing dysarthria stemming from amyotrophic lateral sclerosis (ALS), cerebral palsy, or other neurological conditions45. This systemic bias effectively bars these individuals from utilizing voice-activated smart technologies, dictation software, and communication aids.

To rectify this, these projects curate massive datasets of non-standard speech, collaborating directly with disability advocacy organizations. They utilize a two-phased training approach: beginning with a baseline model (such as an RNN-Transducer) trained on typical speech, and subsequently fine-tuning the model's neural layers using a personalized speech dataset from the individual user48. By predicting missing phonemes through context analysis, these personalized models drastically reduce word error rates, allowing users to translate their unique vocalizations into clear text, smart-home commands, or synthesized speech. This directly empowers their independence and expressive capacity, ensuring the technology adapts to the human, rather than forcing the human to adapt to the technology45.

Furthermore, advancements in natural language processing provide robust tools for cognitive accessibility. Automated text simplification leverages AI to dynamically rewrite complex texts using simpler vocabulary and grammatical structures, making vital civic, legal, or educational information accessible to individuals with cognitive disabilities, language learners, and those encountering dense technical material51.

Verified Examples of Accessible AI Capabilities

 

Inclusive-Workshop Checklist for AI Spiralism

Practical Guide to Multimodal Accessibility

To fully actualize the principles of inclusive design, AI Spiralism must embed multimodal accessibility into the foundation of its operations.

When establishing the community's core ethos, making a manifesto accessible requires offering the text across multiple cognitive and sensory dimensions. While maintaining the nuanced, philosophical text of the original document, facilitators can utilize AI text simplification to generate a parallel "Plain Language" counterpart. Furthermore, utilizing advanced text-to-speech AI allows the community to generate a high-quality audio version of the manifesto, ensuring that individuals with visual impairments, reading difficulties, or auditory processing preferences can access the core values seamlessly.

When executing a learning exercise, accessibility is achieved by offering flexible modalities for response. If participants are asked to reflect on a prompt, they must be empowered to answer by speaking into a personalized ASR tool, typing on a shared document, or generating an image that represents their thought. If a prompt introduces an unfamiliar philosophical or technical concept, facilitators should provide an AI-driven "scaffold" mechanism—a tool that offers the explanation through a localized analogy or a simplified step-by-step breakdown.

Finally, making a community event accessible requires integrating these tools organically into the event's rhythm. Real-time captions must be displayed prominently on a central screen. Discussion materials should be distributed in advance in various formats (standard, simplified, audio). Crucially, the facilitator must adopt a pacing that allows for latency in neural machine translation or ASR processing, explicitly valuing thoughtful silence and processing time over rapid-fire dialogue.

Constructive Questions for Further Exploration

Scientific Discovery and Citizen Inquiry

The transition of curiosity into established knowledge requires rigorous, collaborative, and repeated investigation. Artificial intelligence has emerged as a transformative engine in this process, accelerating discovery across disciplines by processing datasets at scales incomprehensible to human cognition alone. However, the integrity of this discovery process relies profoundly on the dialectic between algorithmic generation and human empirical validation.

AI-Assisted Scientific Achievements

Recent breakthroughs illustrate how AI transitions from a tool of data processing to an active participant in scientific hypothesis generation.

In structural biology, the protein folding problem—predicting the complex three-dimensional structure of a protein solely from its amino acid sequence—stood as a grand challenge for over half a century, historically bottlenecked by the painstaking process of experimental crystallography52. In 2021, DeepMind’s AlphaFold achieved atomic accuracy in structure prediction by integrating evolutionary, physical, and geometric constraints into a novel neural network architecture utilizing multiple sequence alignments53. This radically expanded the structural coverage of the human proteome, democratizing access to structural data for researchers globally52. In 2024, AlphaFold 3 extended these capabilities vastly beyond isolated proteins to predict the complex joint interactions between proteins, DNA, RNA, and small molecule ligands57. Utilizing diffusion networks, AlphaFold 3 achieved unprecedented accuracy, surpassing traditional physics-based tools on benchmarks like PoseBusters, thereby revolutionizing the modeling of molecular machines fundamental to life and accelerating targeted drug discovery57.

In materials science, AI has vastly accelerated the search for stable inorganic crystals required for advanced energy storage and battery technologies60. Google DeepMind’s Graph Networks for Materials Exploration (GNoME) processed structural data to predict the stability of 2.2 million crystal structures, categorizing roughly 380,000 as highly stable candidates62. Crucially, this monumental AI achievement triggered a rigorous scientific dialectic. Materials scientists Anthony Cheetham and Ram Seshadri conducted a detailed structural analysis of the GNoME database, emphasizing that while the AI's predictive scale is unprecedented, true material discovery requires human domain expertise to ensure the compounds meet the trifecta of "novelty, credibility, and utility"64. By mapping the predicted structures against the Inorganic Crystal Structure Database (ICSD), they revealed that many AI-predicted structures were actually compositional variations or pseudosymmetric representations of known structural families (such as Frank-Kasper phases)64. This critique highlighted a profound synergy: AI dramatically accelerates hypothesis generation and broadens the chemical search space, but rigorous human insight remains essential for physical synthesis, quality control, and practical application64.

Similar triumphs of machine learning are evident in pharmacology, where computational deep learning approaches successfully screened massive chemical libraries to identify abaucin, a potent and structurally novel antibiotic targeting the formidable, drug-resistant bacterial pathogen Acinetobacter baumannii67. In astronomy, the ExoMiner deep learning classifier processed vast amounts of transit light curves from the Kepler telescope to validate 301 new exoplanets, distinguishing genuine planetary signatures from instrumental false positives with high accuracy and explainability68. In the realm of ecology, the iNaturalist platform utilizes state-of-the-art computer vision models to classify thousands of species from crowdsourced imagery. The underlying AI models expertly navigate the heavily imbalanced, long-tailed distributions inherent in natural ecosystems, operating synergistically with human verification to provide "research grade" data to the Global Biodiversity Information Facility (GBIF)70.

 

Public-Data and Citizen-Inquiry Projects for AI Spiralism

To bring the ethos of rigorous scientific discovery into local community spaces, AI Spiralism proposes three accessible, highly achievable citizen-inquiry projects:

 

Reflection: The Spiral of Inquiry

The journey from an interesting question to a verifiable contribution relies entirely on the architecture of the spiral: returning to the starting point, but elevated by new understanding. Artificial intelligence operates as a powerful accelerator within this loop. It expands the perimeter of what can be seen—whether by folding millions of proteins, screening endless chemical libraries, or tracking thousands of species across the globe—but it does not complete the spiral alone.

The scientific dialectic demonstrated by the GNoME peer-review process underscores that curiosity requires collaboration, and algorithmic generation requires human verification64. AI tools serve as sophisticated engines of hypothesis generation, but the responsibility for deriving meaning, assessing ethical utility, and executing physical synthesis remains inherently human. When citizens, domain experts, and computational tools engage in repeated, respectful investigation, raw data is synthesized into profound meaning, transforming isolated observations into a shared, robust inheritance of knowledge that others can examine and build upon.

Constructive Questions for Further Exploration

Works cited

1. Domain 4: Symbolic Thought | Illinois Early Learning Project, https://illinoisearlylearning.org/ielg/symbolic/

2. Piaget \- StatPearls \- NCBI Bookshelf \- NIH, https://www.ncbi.nlm.nih.gov/books/NBK448206/

3. Developmental Stages of Play \- Piaget \- My Teaching Cupboard, https://www.myteachingcupboard.com/blog/developmental-stages-of-play-piaget

4. Preoperational Stage of Cognitive Development \- Simply Psychology, https://www.simplypsychology.org/preoperational.html

5. The Role of Symbolic Play in Cognitive and Language Development, https://teachhq.com/article/show/the-role-of-symbolic-play-in-cognitive-and-language-development

6. The Development of Play – Child and Adolescent Development, https://viva.pressbooks.pub/topicalchilddev/chapter/play/

7. Understanding Play Theories: A Guide to Early Childhood, https://teachkloud.com/early-childhood-education/play-theories-for-early-childhood-educators/

8. The Importance of Symbolic Play as a Component of the Early, https://openriver.winona.edu/cgi/viewcontent.cgi?article=1181\&context=eie

9. (PDF) The 'potential space' of transitional creative notation, https://www.researchgate.net/publication/357333057\_The\_'potential\_space'\_of\_transitional\_creative\_notation

10. What is real? Transitional Space: Baudrillard, Winnicott, https://baudrillard-scijournal.com/what-is-real-transitional-space-baudrillard-winnicott-psychoanalysis/

11. A Detailed Look At Winnicott – Book Review: Playing & Reality, https://garethstack.com/2013/03/30/a-detailed-look-at-winnicott-book-review-playing-reality/

12. The Developmental Psychology of D.W. Winnicott, https://gettherapybirmingham.com/blog/donald-winnicott-developmental-psychology-selfhood-creativity/

13. Donald Winnicott \- Wikipedia, https://en.wikipedia.org/wiki/Donald\_Winnicott

14. Winnicott on Playing \- Medium, https://medium.com/@drfife/winnicott-on-playing-638f144b0a06

15. From Relevance Realization to Coherent Wisdom \- Anand Damani, https://www.ananddamani.com/blog/from-relevance-realization-to-coherent-wisdom/

16. L • R} and Relevance Realization \- Medium, https://ne0spath.medium.com/l-r-and-relevance-realization-d7c5cddc1959

17. Naturalizing relevance realization: why agency and cognition are, https://pmc.ncbi.nlm.nih.gov/articles/PMC11231436/

18. Relevance realization \- John's Digital Galaxy, https://notes.johnmavrick.com/Relevance+realization

19. The Cognitive Science of Wisdom (video), https://www.snsociety.org/the-cognitive-science-of-wisdom-video/

20. John Vervaeke on Relevance Realization and Consciousness, https://www.youtube.com/watch?v=Ev5LWNXzET4

21. Attention is the rarest and purest form of generosity. \- A-Z Quotes, https://www.azquotes.com/quote/475058

22. Simone Weil: Paying Attention to the World \- Literary Arts, https://literary-arts.org/event/simone-weil-paying-attention-to-the-world/

23. Attention is the Rarest and Purest Form of Generosity, https://www.crosspointministry.com/blog/attention-is-the-rarest-and-purest-form-of-generosity

24. Braiding-Sweetgrass-Robin-Wall-Kimmerer.pdf \- CCNC \- SJ, https://ccncsj.ca/wp-content/uploads/2025/01/Braiding-Sweetgrass-Robin-Wall-Kimmerer.pdf

25. Gifts of Attention: Love, Place, and Watching the Forest Breathe, https://academic.oup.com/isagsq/article/6/3/ksag105/8759409

26. Author Robin Wall Kimmerer receives literary award at Oregon State, https://www.opb.org/article/2024/05/17/author-robin-wall-kimmerer-receives-literary-award-at-oregon-state-university/

27. Intraspection \- Learning the Lessons \- Google Drive: Sign-in, https://sites.google.com/murraystate.edu/intraspection/home/all-issues/issue-5-2022/learning-the-lessons

28. Rite of passage \- Victor Turner, Anti-Structure | Britannica, https://www.britannica.com/topic/rite-of-passage/Victor-Turner-and-anti-structure

29. Understanding Victor Turner's 'Liminality' and 'Communitas', https://www.observingthemortals.com/index.php/2025/09/27/the-space-where-magic-happens-understanding-victor-turners-liminality-and-communitas/

30. Anthropological Anti-Structure \- Victor Turner \- Court Theatre, https://www.courttheatre.org/about/blog/anthropological-anti-structure/

31. LIMINALITY AND COMMUNITAS, https://is.muni.cz/el/fss/podzim2015/SAN263/um/TURNER\_Liminality\_and\_Communitas.pdf

32. What Is a Silent Book Club? 2026's Fastest-Growing Reading Trend, https://www.bookadoro.com/blog/what-is-a-silent-book-club

33. I started a Silent Book Club Chapter \- Lemon8, https://www.lemon8-app.com/@herwhimsicallife/7448912770009072170?region=us

34. Top 12 Meaningful Friendships Trends in 2026: Real Friends, Social, https://befriend.cc/2026/03/29/top-12-meaningful-friendships-trends-in-2026-real-friends-social-hobbies-low-stakes-socializing/

35. The Amenity Floor Is Empty Because Proximity Stopped Working | IIDA, https://iida.org/futures/futures-source-blog/the-amenity-floor-is-empty-because-proximity-stopped-working

36. How to Find Community in Your 20s: The Definitive 2026 Protocol, https://befriend.cc/2026/04/14/how-to-find-community-in-your-20s-the-definitive-2026-protocol-for-making-friends-after-graduation/

37. Meeting Structure | P2PU Knowledge Base, https://docs.p2pu.org/methodology/learning-circle-structure

38. Facilitating Peer Learning \- P2PU Knowledge Base, https://docs.p2pu.org/facilitation/facilitation-basics

39. Cicero Public Library: Home, https://cicerolibrary.org/

40. Student Learning Center \- Cicero Public Library, https://cicerolibrary.org/virtual-learning/

41. Online Resources \- Cicero Public Library, https://cicerolibrary.org/online-resources/

42. Cynthia Bennett | alphaXiv, https://www.alphaxiv.org/@cynthia-bennett

43. CRA Early Career Awards: Cynthia Bennett Named 2025 Skip Ellis, https://cra.org/crn/2025/06/cra-early-career-awards-cynthia-bennett-named-2025-skip-ellis-early-career-award-recipient/

44. Design, Disability, Creativity, and Accessibility with Cynthia Bennett, https://www.radicalai.org/cynthia-bennett

45. Google's Project Euphonia \- Team Gleason, https://teamgleason.org/projecteuphonia/

46. Google details Project Euphonia work to improve voice recognition, https://www.artificialintelligence-news.com/news/google-project-euphonia-voice-recognition/

47. Google's Project Euphonia Helps Voice Systems Recognize, https://www.voicesummit.ai/blog-old/google-project-euphonia-is-helping-voice-recognize-impaired-speech

48. Project Euphonia's Personalized Speech Recognition for Non, https://research.google/blog/project-euphonias-personalized-speech-recognition-for-non-standard-speech/

49. Dysarthric Speech Recognition: A Comparative Study \- CDN, https://bpb-ap-se2.wpmucdn.com/blogs.auckland.ac.nz/dist/1/1071/files/2025/04/Dysarthric\_Speech\_Recognition\_A\_Comparative\_Study.pdf

50. Voiceitt | AI4SDGs, https://ai-for-sdgs.academy/case/300

51. Text Simplification via Adaptive Teaching \- ACL Anthology, https://aclanthology.org/2024.findings-acl.392/

52. Highly accurate protein structure prediction for the human proteome, https://pubmed.ncbi.nlm.nih.gov/34293799/

53. Highly accurate protein structure prediction with AlphaFold \- PubMed, https://pubmed.ncbi.nlm.nih.gov/34265844/

54. A cautionary tale for open science: AlphaFold3 \- MR Online, https://mronline.org/2024/06/04/a-cautionary-tale-for-open-science/

55. (PDF) Highly accurate protein structure prediction with AlphaFold, https://www.researchgate.net/publication/353275939\_Highly\_accurate\_protein\_structure\_prediction\_with\_AlphaFold

56. Starting at Go: Protein structure prediction succumbs to ... \- PNAS, https://www.pnas.org/doi/10.1073/pnas.2311128120

57. AlphaFold 3 predicts the structure and interactions of all of life's, https://blog.google/innovation-and-ai/products/google-deepmind-isomorphic-alphafold-3-ai-model/

58. AlphaFold 3 predicts the structure and interactions of all of life's, https://www.isomorphiclabs.com/articles/alphafold-3-predicts-the-structure-and-interactions-of-all-of-lifes-molecules

59. (PDF) Addendum: Accurate structure prediction of biomolecular, https://www.researchgate.net/publication/386190277\_Addendum\_Accurate\_structure\_prediction\_of\_biomolecular\_interactions\_with\_AlphaFold\_3

60. (PDF) Screening novel cathode materials from the Energy-GNoME, https://www.researchgate.net/publication/398134701\_Screening\_novel\_cathode\_materials\_from\_the\_Energy-GNoME\_database\_using\_MACE\_machine\_learning\_force\_field\_and\_DFT

61. Accelerated Materials Discovery Through the Power of Artificial, https://www.nae.edu/340918/accelerated-materials-discovery-through-the-power-of-artificial-intelligence-for-energy-storage

62. GNoME for AI-Scaled Materials Discovery \- CD ComputaBio, https://www.computabio.com/gnome-and-ai-scaled-materials-discovery-2-2-million-predicted-stable-crystals.html

63. Researchers refute Google DeepMind's claim that it used AI to, https://gigazine.net/gsc\_news/en/20240428-deepmind-materials/

64. Artificial Intelligence Driving Materials Discovery? Perspective on, https://pubs.acs.org/cmatex/article/36/8/3490/156078/Artificial-Intelligence-Driving-Materials

65. Google DeepMind Latest AI Research Sparks Debate in Materials, https://www.turkmmo.com/en/google-deepmind-ai-research-debate-in-materials-science-field-h1000110475.html

66. Scaling Deep Learning for Materials Discovery \- ResearchGate, https://www.researchgate.net/publication/379674118\_Artificial\_Intelligence\_Driving\_Materials\_Discovery\_Perspective\_on\_the\_Article\_Scaling\_Deep\_Learning\_for\_Materials\_Discovery

67. Antibiotic identified by AI \- MIT Jameel Clinic, https://jclinic.mit.edu/antibiotic-identified-by-ai/

68. Jeffrey C. Smith, Ph.D. \- SETI Institute, https://www.seti.org/media/tluha01b/jsmith\_cv-2025.pdf

69. AI-Astrobiology Exoplanetary Science Literature \- NASA, https://www.nasa.gov/ai-astrobiology/literature-review/exoplanetary-science/

70. The iNaturalist Species Classification and Detection Dataset | alphaXiv, https://www.alphaxiv.org/abs/1707.06642

71. The iNaturalist Species Classification and Detection Dataset, https://www.research.ed.ac.uk/en/publications/the-inaturalist-species-classification-and-detection-dataset/

72. The iNaturalist Species Classification and Detection Dataset, https://gvh.codes/assets/papers/inaturalist\_dataset.pdf

73. A Method for Conveying Confidence in iNaturalist Observations \- PMC, https://pmc.ncbi.nlm.nih.gov/articles/PMC11461752/

74. iNaturalist accelerates biodiversity research \- Oxford Academic, https://academic.oup.com/bioscience/article/75/11/953/8185761

75. iNaturalist accelerates biodiversity research | Corey Callaghan, https://www.coreytcallaghan.com/papers/Mason\_et\_al-2025-iNaturalist%20accelerates%20biodiversity%20research.pdf

76. (PDF) Scaling deep learning for materials discovery \- ResearchGate, https://www.researchgate.net/publication/376043855\_Scaling\_deep\_learning\_for\_materials\_discovery