AI Psychology Video Generator

Stick Figures, the default here, draws hand-drawn doodle characters, the plainest look there is. With nothing in the art to admire, the attention goes to the idea, which suits cognitive biases, habits and 'why do we…' questions told with a straight face and a little humour. It's plainly an illustration, so it can show a feeling, a bias or an old experiment's setup without pretending to be footage. A video in it typically costs about 25 credits.

Whiteboard draws marker doodles on white, the look of a lesson. It suits a concept with parts, one step per scene: what the testing effect is and how to use it, the stages of a model, the design of an experiment and then its result. About 25 credits.

Cartoon Explainer draws bold, expressive cartoon characters, which turn a bias into a small story: a shopper anchored by the first price they saw, a manager who only hears what confirms her plan. The AI plans recurring characters, a shared setting and a palette before any image is drawn; small details can still slip, so describe your main character once near the start of the script. About 25 credits.

Pencil Sketch, graphite on paper, reads like a notebook, which suits a classic experiment told as a reconstruction: Hermann Ebbinghaus testing his own memory on lists built from about 2,300 nonsense syllables for his 1885 book, or a famous study's setup before its result. Flat Vector's clean shapes are the science-explainer look, for how memory, attention or sleep work, where a mechanism is easier to follow as simple shapes than as anatomy. Each typically costs about 25 credits.

Dark Mystery is moody, foggy and low-lit, the look viewers expect from 'dark psychology'. It fits topics such as the Dark Triad, the traits of narcissism, Machiavellianism and psychopathy named in a 2002 paper by Paulhus and Williams, or how manipulation works, as long as the video explains and warns rather than teaches (see below). Its scenes look photographic, which matters for AI labels, and it's the wrong register for suicide or self-harm. About 25 credits.

Documentary stock footage is the cheapest, at about 12 credits a video: real clips of people, crowds and city streets, licensed for commercial use with nothing to attribute. It shows people in general, not the experiment or its participants, so the narration has to carry the finding. Mixed, which combines footage and AI images, typically costs about 14.

Every look above except documentary footage and Mixed can also turn each image into a short AI video clip. That raises a typical video to about 464 credits, so keep it for a series' flagship explainer rather than the daily post.

"Does a habit really take 21 days? In a 2010 UCL study, 96 volunteers repeated one eating, drinking or activity behaviour daily; on average it took 66 days to become automatic, with a range of 18 to 254 days, and missing one day didn't materially affect the process. Stick figures, dry humour, end on why the range matters more than the average."

"Why testing yourself beats rereading. In a 2006 study, Roediger and Karpicke had students study prose passages, then reread them or recall them from memory. Five minutes later, rereaders did better; two days or a week later, the self-testers remembered more, though rereading had made students more confident. Whiteboard look, calm teacher voice, end on one way to test yourself tonight."

"Are people less likely to help when others are around? A 2011 meta-analysis in Psychological Bulletin, pooling bystander studies from the 1960s to 2010 with over 7,700 participants, found they are on average, but less so when the emergency is clearly dangerous. Cartoon Explainer look, a busy station platform, calm narrator, end on why an obvious emergency gets more help."

Each fits the 400-character limit of the box above and names the study, its year and what it found. Write the finding into the prompt yourself rather than asking for 'five psychology facts': then you know what the narrator will say, and you can put the full reference in the post text.

Psychology has a replication problem, and short videos inherit it. In 2015 the Open Science Collaboration, publishing in Science, repeated 100 studies from three psychology journals: 97% of the originals had reported statistically significant results, but only 36% of the repeats did, and the repeated effects were on average half the size. One study, however famous, is a starting point, not a fact.

Power posing is the best-known case. In 2010, Carney, Cuddy and Yap reported in Psychological Science that holding two 'high-power' poses for one minute each raised testosterone, lowered cortisol and increased feelings of power and tolerance for risk. In 2015 a replication in the same journal by Ranehill and colleagues found no effect on hormones or risk tolerance in a large sample of men and women. The first author, Dana Carney, later posted a statement on her Berkeley faculty page calling the original sample of 42 'tiny' and saying: 'I do not believe that "power pose" effects are real.'

Other findings were overstated rather than overturned. Shoda, Mischel and Peake's famous 1990 marshmallow study reported strong links between how long preschoolers could wait for a bigger treat and their achievement as teenagers. A 2018 conceptual replication in Psychological Science by Watts, Duncan and Quan, following children whose mothers hadn't completed college, found a link half that size, and it shrank by two thirds once family background, early cognitive ability and the home environment were taken into account. 'Kids who waited became more successful' is the viral version; 'a small link that mostly disappeared once family background and early ability were accounted for' is closer to the study.

So script what was measured, not what it 'proves'. Name the study, the year and what it measured: 'in a 2006 study, students who tested themselves remembered more a week later than students who reread' rather than 'studies show testing is the best way to learn'. Say how many people took part when the study was small, and what later work found when there is a replication. Then put the reference in the post text: authors, year and journal fit on one line.

Dark psychology needs one more rule: explain, never coach. Write from the side of the person on the receiving end: what a tactic looks like, why it works on people, how to recognise it and where to get help, never a step-by-step for using it on someone. 'The signs of love bombing' is an explainer; 'how to make someone obsessed with you' is coaching. And don't diagnose: a video can explain what narcissistic personality disorder is, but it can't tell a viewer that their partner has it.

YouTube's monetisation policies include 'AI Personas Related to Sensitive Topics'. It covers 'channels that use AI-generated personas to deliver information on sensitive topics', including 'any content that presents itself as a human expert providing advice to viewers on topics such as health, legal issues, finances, or politics', and says 'channels uploading this content will not be allowed to monetize'. Its first example is 'An AI "doctor" providing medical diagnoses, health advice, or wellness remedies'.

For psychology, the line not to cross is an AI narrator presenting as a psychologist or therapist: 'as a therapist, I…', 'my clients', a diagnosis of the viewer, or advice on treating their anxiety. YouTube doesn't define 'persona' further, so give the narrator no expert identity at all, and keep therapists out of the scenes too: the AI plans recurring characters from your script before it draws, so one line about a therapist can give the whole video one. A voice that explains what a 2006 study measured and names its source is a narrator, not a therapist.

Mental health itself is a topic both platforms allow. YouTube says its users 'should not be afraid to speak openly about the topics of mental health, suicide, self-harm, and eating disorders in a supportive and non-harmful way'. Its suicide and self-harm policy doesn't allow content that promotes or glorifies them, gives instructions, or is 'targeted at minors', and it weighs whether content is 'meant to be educational, documentary, scientific, or artistic' when it decides whether a video stays up, is age-restricted or gets a crisis resource panel. Its best practices for creators ask for prevention resources 'in both the video itself and the description', no mention of methods or locations, and: 'Do not use sensationalist language or dramatic visuals.' That rules out Dark Mystery for these topics; Whiteboard or Pencil Sketch is the calmer choice.

On ads, YouTube's advertiser-friendly guidelines list self-harm, suicide and eating disorders among 'controversial issues'. Content 'related to preventing' them can earn ad revenue; an 'educational, documentary, or scientific representation of controversial issues' is listed under limited ad earnings; and an 'animated depiction of controversial issues presented in a sensational manner' earns none.

TikTok's Community Guidelines allow 'information on suicide or self-harm prevention, like warning signs or where to get help', and educational content that 'raises awareness about suicide and self-harm, unless it includes detailed descriptions or imagery'. Educational content with detailed descriptions or non-graphic imagery is ineligible for the For You feed and age-restricted, and showing, promoting or giving instructions for suicide or self-harm isn't allowed.

Both platforms also have rules for health claims. YouTube's medical misinformation policy doesn't allow content that 'contradicts health authority guidance on treatments for specific health conditions'; its named examples are mostly about vaccines and cancer, and it may make exceptions for 'content discussing the results of a specific medical study'. TikTok doesn't allow health misinformation that could cause significant harm, such as 'discouraging professional care for life-threatening conditions', and makes this ineligible for the For You feed: 'selectively referencing certain scientific data to support a conclusion that is counter to the findings of the study'. That is the cherry-picked psychology fact, described exactly.

For AI labels, the look decides. YouTube asks you to disclose AI content that looks realistic and shows a real person doing or saying something they didn't, alters footage of a real event or place, or generates a realistic scene that didn't happen. You answer Yes under 'AI use' when you upload; for videos ShortFast posts for you, the setting is in YouTube Studio. Its examples that need no disclosure include clearly unrealistic content, animation, and using AI for the script or the captions, and it says disclosing alone won't limit a video's audience or its eligibility to earn money. Stick figures, whiteboard doodles, cartoons, sketches and flat shapes are plainly not footage. Dark Mystery looks photographic: when it shows a real case, a real experiment or a real person as if real, answer Yes.

TikTok requires a label on AI-generated content containing realistic images, audio or video; artistic styles and generic text-to-speech narration don't need one. It also encourages labeling anything completely generated or significantly edited by AI, and these videos are made with AI from script to voice, so turn on its AI-generated content setting when you post: TikTok says the setting won't affect distribution. Even with a label, TikTok doesn't allow AI content that shows fake authoritative sources, so never dress a video up as a statement from a university, a journal or a health body.

Finally, a stick-figure or cartoon look doesn't make a psychology video made for kids. YouTube lists 'animated characters or cartoon figures' among its factors, but says content 'intended for a teen or older audience' is general audience and 'should be set as "not made for kids"'. ShortFast doesn't set an audience on the videos it posts, so set it once for the whole channel in YouTube Studio, under Settings, Channel, Advanced settings.

It doesn't check research claims. ShortFast writes a fluent script, and fluent isn't the same as true: a draft can repeat a finding that didn't replicate, inflate a small effect or attach a result to the wrong study. Nothing in ShortFast verifies a study, a number or a quote, whether you paste the script or it writes one, so check each claim against the paper or the university's page before you render.

Stick figures can't show real experiments. Every scene is a new drawing made from your script, not footage of the study, its lab or its participants, so present it as a reconstruction: 'here's how the experiment was set up', not 'here's what happened'. If you have the right to use real photographs from a study, upload them into the scenes.

The narration is fixed once rendered. You can edit captions, swap scenes, change music, title and post text and re-render, but a wrong year or an overstated result in the narration means generating again. There are no on-screen citations either: the only text ShortFast adds is the burned-in captions of what the narrator says, so say the study's name and year out loud and put the full reference in the post text.

An automation has no approval step. Each episode in a series posts automatically or arrives by email, with a topic and a script the AI wrote, so for psychology choose email delivery and check every study it names before you post. The next two episodes are queued: you can edit their captions, scenes and post text, re-time, pin or delete them, but not change their script.

None of these channels use ShortFast; they show what psychology audiences reward. Psych2Go, whose channel describes a mission of 'making psychology, mental health, and self-discovery accessible', narrates relatable questions such as 'Why You Can’t Stop Thinking About Someone' over a cast of cartoon characters. Sprouts ('We create cartoons that explain complicated ideas in simple language') draws simple hand-drawn figures on flat colour backgrounds, the look closest to Stick Figures, for concepts and experiments such as 'Correlation vs Causation Explained: Why Patterns Can Mislead Us' and 'How Power Corrupts [The Rigged Monopoly Experiment]'. Practical Psychology explains one concept or study per video over illustrated, animated scenes, with titles such as 'The Asch Line Study - Conformity Experiment' and 'Experimenter Bias (Definition + Examples)'. The School of Life narrates essays on emotional life, such as 'What Do I Project onto Other People?', over collage-style illustrations.

What they share is one idea per video, named plainly in the title, and a look you would recognise with the sound off. The idea is the part to copy; the scenes, the voice and the captions are the part ShortFast handles.