The MMA Training Bible is a one-person operation. I’m Dr. Jason Gillis. I’ve worked in MMA sport science since 2008, and I’ve fought as an amateur myself. Everything here, including the videos, the articles, and the writing behind them, is built and owned by me. What you’re reading is my own understanding of the science, updated against current research and written for fighters. I use AI tools in the process, and I’d rather tell you exactly where than let you guess.
Here’s how it works. The argument of every section is framed by me before any AI touches it. I build every section’s claim, its evidence, and its limits and when I can’t, that’s my cue to go back to the papers. From there, AI helps me draft and organize the material, critically analyze the structure of an idea from both a coach’s and a scientist’s point of view, turn dense research into readable prose, and shape finished writing. Where a draft sounds too complex, or just doesn’t sound right, I have it redone in plain language, with analogies that actually teach, and then I tweak that wording myself until it says what I mean, the way I’d say it, based on my own understanding of the material. The version you read is the one I shaped and signed off on, line by line. AI fills in the middle. The thinking at both ends is mine.
So, AI speeds up the slow parts of writing. It does not decide what counts as evidence, and it does not decide what’s true. The obligations stay with me, at three specific points:
I choose the research. I run the literature searches myself and decide which studies are good enough and relevant enough to include, using the same screening method I’d use for a systematic review. AI doesn’t pick my sources.
I check the claims against the papers. Where a draft says more than the study found, i.e. a cautious finding stated as a sure thing, or a stretched result, I correct it. Overstatement and hallucination are the main ways assisted writing goes wrong, and they’re what I search for.
Every number you see is traced by hand to the study it came from. I read the paper, find the figure in the results, and confirm it says what my draft claims, both in value and in context. When it doesn’t, I fix it.
There’s a principle underneath all of that: the line that matters isn’t between assisted and unassisted writing. It’s between expression and obligation. Expression is the wording: how an idea gets phrased, organized, made readable. Obligation is what I owe you: that the claims are true, the citations are real, the numbers match their sources, and that when something’s wrong, I correct it. AI can help with expression. The obligations can’t be delegated, and I don’t try.
This is, incidentally, the same standard the major scientific publishers have converged on for AI-assisted work: disclosed assistance, genuine intellectual contribution, and a named human fully accountable for every claim. I hold this site to it too.
One more thing I want to mention. Writing isn’t just output. Working through the prose is part of how you think. It’s where arguments get tested and where you find out what you don’t actually understand. I take that seriously, and it’s why the initial framing and the final pass are the parts I keep.
One honest limit. All of this makes sure I report the research accurately, it doesn’t make any one study the final word. Science moves, single findings get overturned, and some questions are still open. When that’s the case here, I say so.
So: assisted in the expression, accountable for the obligations. If you ever find a number here that doesn’t match its source, tell me — I’ll correct it and thank you for it.
Last updated: August 2026.