The AI statements have been copied directly from sources such as webpages and docs which have been cited accordingly. No edits have been made to the texts. An inclusion on this page is not an endorsement. That said, I am a fallible human being with biases, so you may see more of one perspective than the other. That is why, please send me any and all AI statements you find so that I can include them here for a broader perspective.
I have tried my best to ask for permission to include statements. In case your statement is here and you want it removed, kindly let me know.
AI and LLM use: I follow the first two membership commitments of the Association for Human Mathematics. Thus, I do not work with commercial AI companies, and I do not use AI or LLM tools to generate mathematical texts. I believe that writing should have human authorship.
Referee and review activities: I will only agree to referee submitted manuscripts that include a transparent disclosure regarding AI and/or LLM use/non-use. This might be a disclosure statement contained within the manuscript itself, or it might be a statement contained in the cover letter for the submission. If the AI disclosure statements make clear that the authors did most of the work, and that the AI/LLM was used for what I consider to be minimal assistance (grammar checks, literature searches, etc.), then I will consider accepting the referee or review request. I will decline to referee or review any paper that is produced with AI or LLM use that replaces human authorship and reasoning (AI-generated text, proof arguments, etc.).
I have chosen not to rely on artificial intelligence as a source of novel ideas in my own research. The spectacular progress of artificial intelligence opens unprecedented opportunities to amplify the reach and applications of our discipline. Yet, whether I am exchanging ideas with fellow mathematicians, teaching, mentoring students, or sharing mathematics with the wider public, it is not the answer itself but the path that leads to it which plays the first role. I therefore wish to remain, in some sense, an artisan mathematician, taking the time to wander, alongside colleagues, through the hidden corners of the mathematical landscape.
Also see: https://proofsandprompts.com/2026/08/30/care-for-a-little-more-ai
I am a member of the AI-free caucus in the Association for Human Mathematics. This means that I do not make use of LLMs or other genAI tools in any part of my research or other professional activities. I ask that my co-authors also avoid use of these technologies while working on our joint projects.
I believe that AI is harmful to both humanity and mathematics. For this reason, I do not use artificial intelligence in my work and have no plans to do so in the future. I am also a member of the Association for Human Mathematics.
No portion of any script will be written, edited, or outlined by an LLM.
The thesis of a video will always originate with a human.
No image or music in a video will be generated by AI. If something is accidentally included, best efforts will be made to remove it.
LLM outputs are not trusted as a source.
In the past 15 years of my time in academic research, I have witnessed first-hand the development of machine learning algorithms, both image classification networks and large language models. I am involved in research projects that use neural networks for astronomical data classification and anomaly detection. I am therefore very much aware of the abilities and limitations of machine learning models, more colloquially known as AI.
I believe that AI is a very useful tool when used responsibly, and share in the belief that the AI industry should be regulated.
I think it’s important to be transparent about any usage of AI in my video production process for subscribers. This document therefore exists for me to clearly define for my subscribers and hired production team (see below), where I do and don’t think it’s OK to use AI in my video production process.
I have based the definitions below on two AI documents which I have found useful as guidance:
The STSci document specifically outlines three questions that they encourage people to ask before using any generative AI tools: 1) Would I stake my professional reputation on this? 2) Do I have the expertise to critically evaluate this output? 3) Does this give me the appropriate ownership level of the work?
It’s with these three questions and the University’s policy in mind, that I have incorporated the use of AI tools into my video creation process in the following ways.
The places my team and I will never use AI are for the things you see and hear in videos, or any content produced. That includes:
Writing of scripts (i.e. the actual words I will say in a video, see below for script research)
Producing B-roll footage or graphics for a video (unless it’s generating or debugging code to produce a python animation of e.g. a graph, see below)
Creation of a video thumbnail or any other channel artwork
Creation of merch designs
The places I think it’s OK to use AI when my team and I make videos:
Analysing expected performance of scripted intros/titles/thumbnails to help maximise audience retention or click-through-rate (CTR, e.g. using tools provided in YouTube Studio)
The generation of visual examples to explore thumbnail ideas if words/sketches/mockups fail during the thumbnail creation process (not generation of thumbnail itself, or artwork for thumbnail)
Generation of ideas for titles for a video to optimise CTR/performance
Video topic ideas (e.g. to find where there’s a niche that hasn’t been covered by others, using tools provided in YouTube Studio)
Providing summaries of academic research papers
Help finding research papers as sources for a fact found online (e.g. a fact/number stated on NASA website/blog/press release without a citation)
Explanations of concepts unfamiliar in research papers (with a necessary follow up verification of sources)
Code generation and debugging (most often Python, e.g. for visualising data or producing toy simulations and graphics used as B-roll in videos)
Generation of captions for videos (most often used for captioning short form video with mCaptionsAI tool for Final Cut Pro), which are then verified by a human
AI tools available in video editing programs, such as Final Cut Pro (e.g. to search library footage by keyword, or to auto-generate vertical format clips from a horizontal edit)
As more AI tools become available and are integrated into everyday apps, this will inevitably be a non-exhaustive list. However, this is a working document, and so can be updated. If it is updated, the date at the top of this document will change to reflect that.
Finally, I share people’s concerns over learned societal biases in LLMs, theft of intellectual property, and deepfakes. I am also particularly concerned about the environmental impact of generative AI models due to the large amounts of energy and water for cooling needed for the computational servers. This is an issue that is also applicable to all large web services. I can only hope this problem will be alleviated by the switch to renewable energy sources by countries around the world in the near future. For example, GreenPT data centres are powered 100% by renewable energy. In the meantime, I will ride the generative AI wave with you all with a cautiously critical eye.
The Onion values the creativity and skills of people over machines, despite their unreliability and penchants for compensation, benefits, and breathable air. Generative AI tools used in art, video, and writing are ethically wrong and an existential threat to the livelihood of creative professionals. AI is also a detriment to our audience, whom we reluctantly acknowledge. AI generated art, video, and writing will not be used in any creative endeavor at or affiliated with The Onion.
Again, we want to stress that there will be absolutely no use of AI on our site. Not in the writing, not in the editing, not in the images, nothing. If people don’t take a stand against this, well, we’re left with nothing but bland regurgitated drivel, and that isn’t good for anyone!
From this Google doc
For good reason, AI tools such as ChatGPT, Claude, DALL•E 2, and Midjourney have generated a lot of interest. Their ability to synthesize stories and images can be impressive. But they're also incompatible with much of what we do here at Complexly. While we don’t want to ignore the role they might legitimately play in brainstorming and early research, it’s important to note that using these tools to create scripts and visuals often conflicts with Complexly’s values in a number of ways:
The integrity of our content distinguishes it, and these tools currently make a LOT of mistakes. Often, they confidently make things up. We have fact checking processes in place, but it’s very hard to catch errors at the rate these tools produce them.
We respect our fellow creators and our audience. These new tools cannot replace human creativity. Because they’re often trained on people's intellectual property without permission, using scripts written by ChatGPT or illustrations made by DALL•E 2 is not consistent with our respect for each other.
We imagine the world complexly, and currently these tools are not capable of this. ChatGPT, for example, can not generate original ideas or reflect on what biases it might have.
We need to be confident in the ownership of our content, and legal questions remain about who owns the work produced by these tools.
For these reasons, Complexly’s policy on AI is:
We will not produce videos written, edited, or fact-checked by AI.
This includes not just scripts written in their entirety by these tools, but also those with significant portions composed, edited, or fact-checked by them.
We will not produce videos (or any other public-facing content) containing illustrations or animations created by AI trained on non-consensual data sets.
This includes DALL•E 2 and Midjourney. Some tools are trained on images created by
people who agreed to be in the data set, and those may be used.
When using AI tools trained on consensual data sets to create visual elements, we will ensure we are not misleading viewers as to what is real, or contributing to misinformation. Where applicable, we will label images created this way.
We may use AI tools to suggest ideas (for script subjects, examples, visual elements, etc.) and do preliminary research.
Ideas come from many places in the early stages of conceptualizing and producing a video, and it’s fine for AI to be one of them. Similarly, using these tools to identify sources is acceptable as long as those sources are verified by our team.