Aidar Akmaev
Disclosure: NOXI publishes this guide and is one of the tools it mentions; the method works in any presentation software.
Quick answer: An artificial intelligence presentation should answer four questions in order: what AI is, what it does, what it gets wrong and who sets the rules. For a 10-minute class talk, use 12 slides: a definition, types, how generative AI works, examples, three sourced 2026 facts, risks, rules, what comes next, a conclusion and sources.
Most presentations on AI open with a robot picture and close with "AI will change everything". The ones teachers and managers remember give one definition, three examples the audience already uses, one risk with a number and one law with its date. The rule in one line: define it, show it, weigh it, source it.
In short
- Outline12 slides, 540 seconds planned (9 minutes) plus a 1-minute buffer for a 10-minute slot.
- DefinitionThe OECD's, updated in November 2023: a machine-based system that infers from its input how to generate outputs.
- Facts slideThree 2026 AI Index figures, each with its year and a source line.
- Rules slideEU AI Act: in force 1 August 2024, applicable 2 August 2026; NIST AI RMF (US, voluntary): released 26 January 2023.
- VersionsClass 6–10: 8 slides in 5 minutes. College: 12 in 10. Work intro to generative AI: 6 in 5.
Key numbers
What is artificial intelligence, in simple words?
Artificial intelligence is a computer system that learns patterns from data and uses them to produce an answer: a prediction, content, a recommendation or a decision. The OECD's definition, updated in November 2023, is the one to put on a slide [5].
Three words carry the idea and make a good second slide: infers (it works out an answer from patterns rather than following a rule for every case), input (text, images, sound or numbers go in) and outputs (a prediction, content, a recommendation or a decision comes out).
The OECD says it updated the wording to stay aligned with policy work internationally, including in the European Union and Japan [5], so it holds up in a question-and-answer session better than a definition you wrote yourself. For 11-year-olds, say: "AI is a computer program that learns from lots of examples, then makes a guess about something new."
Types of AI: narrow and general, machine learning, deep learning and generative AI
The types of AI that belong on a slide are narrow AI (built for specific tasks: the AI in today's products), general AI (a hypothetical system as capable as a person across tasks), and three ways of building AI that nest inside each other: machine learning, deep learning and generative AI.
| Type | What it means | Example for the slide |
|---|---|---|
| Narrow AI | Built for a specific task or family of tasks | A spam filter, a route planner, a photo app that groups faces |
| General AI | A system as capable as a person across almost any task; a research goal and a debate | No agreed example; keep it for the future slide |
| Machine learning | Learns patterns from data instead of following hand-written rules | A spam model trained on millions of labelled emails |
| Deep learning | Machine learning with many-layered neural networks | Speech-to-text on a phone; reading a medical scan |
| Generative AI | Creates new text, images, audio, video or code | A chatbot drafting an email; an image generator |
| AI agents | Generative AI that takes steps on a computer to finish a task | An assistant that books a meeting across three calendars |
Draw the nesting as circles inside circles: AI contains machine learning, which contains deep learning, which contains generative AI. The picture answers "is a chatbot the same as AI?" before anyone asks.
Machine learning presentation
A machine learning presentation explains one idea: a system learns a pattern from labelled examples and applies it to new input. Use the spam filter as the running case, then show one error and why more varied examples reduce it.
What should a presentation on AI include?
A presentation on AI should include one definition, three examples the audience already uses, one field in depth, three sourced numbers, one risk with a number, one rule with a date and a conclusion that answers the question you opened with. Those seven elements answer the four questions in the order an audience asks them.
If you are free to pick the angle, phrase your title as a question ("Should a small business use AI in 2027?") and make the last content slide the answer; a question gives the audience a reason to follow 12 slides. For other shapes, see how to structure a presentation.
Artificial intelligence presentation outline: 12 slides in 10 minutes
A 10-minute artificial intelligence presentation fits 12 slides planned at 540 seconds (9 minutes), which leaves a 60-second buffer for a late start or one question. The seconds below are a working budget; move time between slides, but keep the total at 540.
| Slide | What goes on it | What you say | Seconds |
|---|---|---|---|
| 1 | Title and the question you will answer | "By the end, you will know whether [question]." | 20 |
| 2 | What AI is: the OECD definition in one sentence | Read the definition, then the three key words: infers, input, outputs | 50 |
| 3 | Types: narrow vs general; machine learning, deep learning, generative AI | Walk the nested circles from outside in | 60 |
| 4 | How generative AI works in three steps: data, patterns, predictions | One analogy, such as finishing a sentence after reading many books | 60 |
| 5 | AI in daily life: three examples the audience used this week | Ask for a show of hands on one of them | 50 |
| 6 | AI in one field (healthcare, education or business) | One use, one benefit, one limit | 60 |
| 7 | The numbers: three facts from the 2026 AI Index with source lines | Each number once, with its year | 60 |
| 8 | What AI gets wrong: errors, bias, incidents (362 in 2025) | One concrete kind of error, then the number | 60 |
| 9 | Who sets the rules: EU AI Act dates; NIST AI RMF | Two dates and one example of a banned or high-risk use | 50 |
| 10 | What to watch next: two dated trends | "Next year, look for…" | 40 |
| 11 | Conclusion: your answer to slide 1's question | One sentence, then the reason | 30 |
| 12 | Sources | Leave it on screen during questions | 0 |
| Total | 20 + 50 + 60 + 60 + 50 + 60 + 60 + 60 + 50 + 40 + 30 + 0 | 540 s = 9:00 |
The middle is where the substance is: slides 3 to 8 hold 60 + 60 + 50 + 60 + 60 + 60 = 350 of the 540 seconds, or 64.8%. For 15 minutes, multiply each slide's seconds by 900 ÷ 600 = 1.5 rather than adding slides: 540 × 1.5 = 810 seconds planned, 90 spare. Open slide 1 with the question and one thing the audience did today ("Before you got here, at least three AI systems made a choice for you"); our guide to how to start a presentation has more 20-second openers.
Artificial intelligence PPT in 10 slides
If the assignment says "an artificial intelligence PPT of 10 slides", merge slide 4 into slide 3 (types, with one line on how generative AI predicts) and move the sources onto each slide's source line instead of a final slide. You keep all four questions and the plan becomes 540 − 60 = 480 seconds, or 8 minutes.
Generative AI presentation: how generative AI works, explained for a class
Generative AI works in three steps: it is trained on a very large collection of text, images or other data, it learns the patterns in that data, and, given a prompt, it predicts the most likely next piece again and again until the answer is complete. That is why it can sound fluent and still be wrong.
Put the three steps on the slide as data, patterns, predictions, and add one line under the last: nothing in the loop checks the answer against reality. The classroom analogy: "Read every recipe ever written, then write a new one. It will look exactly like a recipe; whether the cake rises is another question." Then the fact that makes the point: the top model read analogue clocks correctly only 50.1% of the time, per the 2026 AI Index [1].
AI in daily life: examples that work in front of a class
The best daily-life examples of AI are ones the audience used in the last 24 hours without noticing: a spam filter, a recommendation, a map route, a phone camera, a translated page or a chatbot. Choose three and ask for a show of hands on one.
| Example | What the AI does | Input → output |
|---|---|---|
| Spam filter | Classifies each email | Email text → "spam" or "inbox" |
| Recommendations | Ranks what to show next | What you watched → next video |
| Maps | Predicts journey times | Traffic data → fastest route |
| Phone camera | Adjusts the image as you shoot | Raw image → sharper photo |
| Chatbot | Generates an answer | Your question → a paragraph ("How would you know if it was wrong?") |
The EU AI Act names spam filters (with AI in video games) as minimal-risk examples with no specific rules, and chatbots as a transparency-risk example: users must be told they are talking to a machine [2].
AI in healthcare, education and business: a slide plan for each
A field slide works when it shows one use, one benefit, one limit and one source, in that order. Healthcare, education and business each have an example the EU AI Act names as high-risk, which gives you a limit with an official source [2].
AI in healthcare presentation
For an AI in healthcare presentation, pick one use the audience can picture (reading scans or robot-assisted surgery), say what it helps with and who stays responsible. The EU AI Act lists robot-assisted surgery among its high-risk examples, permitted subject to strict obligations [2]. Take any health number from a health ministry or hospital source: this is an awareness topic, not clinical advice.
AI in education presentation
For an AI in education presentation, use three linked findings from the 2026 AI Index: over 80% of US high school and college students use AI for school tasks, only half of middle and high schools have AI policies, and 6% of teachers say those policies are clear [1]. Add the EU example: exam-scoring tools are high-risk under the AI Act [2]. For sourced education numbers by country, see our AI in education statistics.
AI in business presentation
For an AI in business presentation, open with adoption (88% of organisations, per the 2026 AI Index [1]), show three tasks (answering customer emails, forecasting stock, drafting product descriptions), then one limit: CV sorting and credit scoring are high-risk examples in the EU AI Act [2]. Close with a recommendation the audience could act on: "use it for drafts, keep a person on every decision about people."
AI facts and statistics for 2026: the numbers slide
The most citable AI statistics for a 2026 presentation come from the 2026 AI Index Report by the Stanford Institute for Human-Centered AI (Stanford HAI): 88% organisational adoption, generative AI at 53% population adoption within three years, 362 documented AI incidents in 2025 and US private AI investment of $285.9 billion in 2025 [1]. Under each one, write the same short source line: "Source: Stanford HAI, 2026 AI Index".
Two more figures from the report: 4 in 5 university students now use generative AI, and 73% of experts vs 23% of the public expect a positive impact on how people do their jobs [1]. Incidents: (362 − 233) ÷ 233 = 129 ÷ 233 = 55.4% more in 2025 than in 2024. Investment: $285.9 billion ÷ $12.4 billion = 23.06, which matches the report's "more than 23 times" [1]. Expert and public views: 73% − 23% = 50 percentage points, the gap the report itself names [1].
Three numbers are enough for one slide. A business talk takes 88%, $285.9 billion and 362; an education talk takes 4 in 5, the 6% of teachers who find school AI policies clear, and 362; a talk on trust takes 73% vs 23% and the finding that trust in government to regulate AI is lowest in the US at 31% and highest in Singapore at 91% [1].
AI ethics presentation: bias, privacy, incidents and jobs
An AI ethics presentation should cover errors, bias, privacy and jobs with one example each, then one documented number, such as the 362 AI incidents of 2025, and one rule that responds to it.
The OECD published these definitions in May 2024 [6]; show the definition next to the count so the audience knows what is counted.
| Issue | What to show | Evidence with a source |
|---|---|---|
| Errors | A fluent answer that is wrong | The top model read analogue clocks correctly 50.1% of the time (2026 AI Index) [1] |
| Bias | A decision about people made from skewed data | CV sorting and credit scoring are high-risk uses under the EU AI Act [2] |
| Privacy | What happens to what you type in | Untargeted scraping for facial-recognition databases is a banned practice [2] |
| Incidents | The trend | 362 documented incidents in 2025, up from 233 in 2024 (2026 AI Index) [1] |
| Jobs | Who expects what | 73% of experts vs 23% of the public expect a positive impact on how people do their jobs [1] |
Say "documented" with incidents: the count measures recorded events, which rise with use and with reporting, not the total harm AI causes.
Who sets the rules: the EU AI Act timeline and the NIST AI RMF
The two rule sets most often cited in AI presentations are the EU AI Act, a law that entered into force on 1 August 2024 and became applicable on 2 August 2026 with some exceptions, and the US NIST AI Risk Management Framework, a voluntary framework released on 26 January 2023 [2][3]. One is binding in the EU; the other is guidance.
| Date | What applies (EU AI Act) |
|---|---|
| 1 Aug 2024 | The Act entered into force [2] |
| 2 Feb 2025 | Prohibited practices and AI literacy obligations [2] |
| 2 Aug 2025 | Governance rules and general-purpose AI obligations [2] |
| 2 Aug 2026 | The Act applicable, with some exceptions; enforcement by the AI Office and Member State authorities; transparency rules from August 2026 [2] |
| Dec 2026 | Prohibition 9: AI systems that generate non-consensual sexually explicit content or child sexual abuse material, introduced as part of the AI Omnibus [2] |
| 2 Dec 2027 | High-risk rules for Annex III uses such as biometrics, education and employment [2] |
| 2 Aug 2028 | High-risk rules for products under Annex I, such as lifts and toys [2] |
The Commission's page says the Annex III high-risk rules "have been extended to 2 December 2027" as a result of the political agreement on the AI Omnibus, a legislative proposal adopted on 19 November 2025, agreed on 7 May 2026 and in force since 27 July 2026; products such as lifts or toys have a transition period until 2 August 2028 [2]. Older slides may show other dates, so check any date you reuse; the page we used was last updated on 19 October 2026. On the slide, write the law as three lines: what it is, two dates and one example the audience cares about ("From 2 December 2027, exam-scoring AI in schools must meet high-risk rules").
The EU AI Act's four risk levels
The EU AI Act sorts AI into four risk levels: unacceptable (banned), high (allowed with strict obligations), transparency (allowed with disclosure duties) and minimal (no specific rules).
| Level | What the Act does | Examples on the Commission's page [2] |
|---|---|---|
| Unacceptable | Prohibits nine practices | Social scoring; emotion recognition in workplaces and education institutions; untargeted scraping for facial-recognition databases |
| High | Permits, subject to strict obligations | AI safety components in critical infrastructure such as transport; exam scoring; robot-assisted surgery; CV sorting; credit scoring |
| Transparency | Permits, with disclosure duties | Chatbots must say they are machines; deepfakes must be labelled |
| Minimal | No specific rules | AI in video games; spam filters |
To score an organisation's own AI risks by likelihood and impact, use the approach in our risk management presentation guide.
The NIST AI Risk Management Framework
The NIST AI RMF is a US framework "intended for voluntary use" that helps organisations build trustworthiness into how they design, develop, use and evaluate AI [3]. Its core has four functions: govern (cross-cutting, through all AI risk management), map (the context that frames risks), measure (quantitative, qualitative or mixed methods) and manage (resources for the risks mapped and measured) [4]. NIST released a Generative AI Profile (NIST-AI-600-1) on 26 July 2024, a concept note for a critical-infrastructure profile on 7 April 2026, and says the framework is being revised as part of the White House AI Action Plan [3].
Future of AI presentation: what to watch next
A future of AI slide should name two trends with dates and a source, not predictions: for example AI agents completing computer tasks, the EU's high-risk rules from 2 December 2027, or the revision of the NIST AI RMF [1][2][3].
| Trend | The dated fact | Why it matters to the audience |
|---|---|---|
| Agents | On the OSWorld benchmark of computer tasks, agent success rose from 12% to about 66% (2026 AI Index) [1] | AI that acts on a computer, not only answers |
| High-risk rules | EU high-risk rules for Annex III uses from 2 December 2027 [2] | Schools and employers using AI on people face stricter duties |
| US guidance | The NIST AI RMF 1.0 is being revised [3] | Frameworks that companies follow will change |
AI topics for presentation: 20 artificial intelligence presentation topics, with first and last slides
A good AI presentation topic is a question about one field that has a source you can cite and an answer you can defend in one sentence. The 20 topics below each come with an opening slide that sets the question and a closing slide that answers it.
| # | Topic | First slide (the question) | Last slide (the answer) |
|---|---|---|---|
| 1 | What is AI, really? | Three phone apps: which are AI? | The OECD definition, and the answer |
| 2 | How a chatbot writes | "How did it know what to say?" | It predicts; it does not look answers up |
| 3 | AI in healthcare | Would you let AI read your X-ray? | Yes, with a doctor signing off |
| 4 | AI in education | 80%+ of US students use AI for school. Now what? | Rules: 6% of teachers call school AI policies clear |
| 5 | AI in business | Should a small business use AI in 2027? | For drafts and forecasts, with a person checking |
| 6 | AI ethics | Who is responsible when AI is wrong? | Your answer: who deploys it, who checks it |
| 7 | Bias in AI hiring | A CV rejected by software: fair? | Why CV sorting is high-risk in the EU |
| 8 | Deepfakes | Real or generated: can you tell? | Labelling duties under the EU AI Act |
| 9 | AI and privacy | What happens to what you type in? | Three questions to ask before pasting |
| 10 | AI and the environment | How much computing does AI need? | The US hosts 5,427 data centres (2026 AI Index) |
| 11 | AI in agriculture | Can a camera spot a sick plant? | One use, one limit, one source |
| 12 | AI in transport | Would you ride in a car with no driver? | Safety components are high-risk in the EU |
| 13 | Machine learning basics | How does a spam filter learn? | Examples in, pattern out |
| 14 | Generative AI and creativity | Is an AI image art? | Your answer, both sides stated |
| 15 | AI agents | What if AI could use your computer? | Agent success on OSWorld: 12% to about 66% |
| 16 | AI and jobs | Experts 73%, public 23%: who is right? | What changes, for whom, by when |
| 17 | The EU AI Act | Which AI is banned in Europe? | Four risk levels and two dates |
| 18 | AI and misinformation | Would you share this headline? | Three checks before you share |
| 19 | AI in sport | Should software make the call? | Where a human referee must stay |
| 20 | The future of AI | What will AI do by 2028? | Two dated facts, not predictions |
Topics 1, 2, 9, 13 and 18 suit class 6–10; 3 to 8, 10, 15 and 17 suit college; 5, 9, 16 and 18 adapt to a work meeting. For "generative AI topics", use 2, 8, 14 and 15.
Artificial intelligence presentation for students and for work: four versions
An AI presentation for students in class 6–10 needs 8 slides in 5 minutes; high school and college take 12 slides in 10 minutes; a college seminar on one field takes 15; a work introduction to generative AI needs 6 slides in 5 minutes.
| Audience | Slides | Time | What changes |
|---|---|---|---|
| Class 6–10 / middle school | 8 | 5 min | Simple words, one example per slide, no investment figures |
| High school / college | 12 | 10 min | The full outline above |
| College seminar on one field | 12 | 15 min | Slides 6–8 cover use, numbers and risk in your field; each slide's seconds × 1.5, so 540 × 1.5 = 810 planned |
| Work: intro to generative AI | 6 | 5 min | The six slides below |
AI presentation for class 6–10: 8 slides in 5 minutes
Plan 270 seconds and keep 30 spare: 1 the question (20 s), 2 what AI is in simple words (40 s), 3 AI in your day (50 s), 4 how it learns from examples (40 s), 5 one surprising use (40 s), 6 what it gets wrong (40 s), 7 one rule (30 s), 8 your answer (10 s). Check: 20 + 40 + 50 + 40 + 40 + 40 + 30 + 10 = 270, plus 30 = 300 seconds. Show things students own (a phone, a game, a map), not robots.
Introduction to generative AI at work: 6 slides in 5 minutes
An introduction to AI presentation at work is a briefing, not a lecture: 1 what generative AI is (40 s), 2 three uses in our team (60 s), 3 what not to paste in (50 s), 4 our policy in three lines (50 s), 5 check every output before it leaves the team (40 s), 6 who to ask (30 s). Check: 40 + 60 + 50 + 50 + 40 + 30 = 270 seconds, with 30 for a question. In the EU, slide 4 can note that the AI Act's AI literacy obligations have applied since 2 February 2025 [2]. For a full session with exercises, use our training presentation guide.
How to make a presentation about AI
To make a presentation about AI, write the question slide 11 will answer, pick one field, collect three numbers from one primary source with their years, build the 12 slides from the outline with action titles ("Documented AI incidents rose 55.4% in 2025", not "Incidents"), put a source line under every number, rehearse against each slide's seconds, and prepare two answers: "Will AI take our jobs?" and "How do you know that number is right?"
How to cite AI facts on slides
Cite an AI fact on a slide with a short source line under the number (organisation, report and year) and the full link on your sources slide. Every figure needs its data year, because AI numbers date quickly: "362 documented AI incidents in 2025" is a fact, "362 AI incidents" is not.
- One number, one source, one year. Never merge two measures from two reports in one sentence; show them side by side, labelled, as in the worked example below.
- Keep the source's words. "Documented incidents" must not become "362 AI disasters".
- Round once, at the end: 129 ÷ 233 = 0.5536, shown as 55.4%.
- Primary source, no invented figures. Cite the report, not the news site that repeated it; if a claim has no number, make it without one.
Tools that research and cite as they build slides save time, but check each number against the page it came from; see our guide to AI presentations with research and citations.
Common mistakes in AI presentations
The five most common mistakes in AI presentations are invented statistics, numbers without years, robot clip-art, treating one chatbot as all of AI, and absolute claims such as "AI will replace all jobs".
| Mistake | Why it hurts | Fix |
|---|---|---|
| Invented statistics | One wrong number undoes the talk in the Q&A | Three numbers from one primary source, each with a source line |
| Numbers without years | AI figures date within a year | "362 in 2025 (2026 AI Index)" |
| Robot clip-art | Says nothing about how AI is used | Show a phone, a map, a scan or a spam folder |
| One chatbot = AI | Confuses a product with a field | The nested-circles slide: AI, ML, deep learning, generative AI |
| Absolute claims | No source supports "all jobs" or "every disease" | A dated fact: high-risk rules from 2 December 2027 |
Worked example: Arjun's "AI in Business" presentation, 10 minutes
1. The question and the field. Arjun uses the 12-slide outline with business on slide 6: customer emails, stock forecasts and product descriptions for a small online shop. His slide 11 answer: "Yes, for drafting and forecasting, with a person checking every output and a written policy."
2. A class survey, labelled as one. He asks: "Did you use a generative AI tool at least once in the past seven days?" 35 of 42 classmates say yes: 35 ÷ 42 = 0.8333, or 83.3%. The 2026 AI Index reports that 4 in 5 university students now use generative AI, or 80% [1]. The two measure different things (past-week use vs use), so slide 7 shows them side by side, each labelled ("Our class survey, 42 students, January 2027"; "Stanford HAI, 2026 AI Index"), without calling one proof of the other. One classmate is 1 ÷ 42 = 2.4 percentage points, so the class figure is a snapshot. Our survey results presentation guide covers reporting a small survey honestly.
3. Facts, risk and rules. Slide 7 carries 88% organisational adoption, US private AI investment of $285.9 billion in 2025, and his class figure next to the Index's 4 in 5 [1]. Slide 8 shows documented AI incidents as a pair of bars, 233 in 2024 and 362 in 2025 [1]: 362 − 233 = 129, and 129 ÷ 233 = 0.5536, a rise of 55.4%; it adds the business limit that CV sorting and credit scoring are high-risk uses in the EU [2]. Slide 9 gives two dates, applicable since 2 August 2026 and high-risk rules for uses such as employment from 2 December 2027 [2], and one line on the voluntary NIST AI RMF [3].
4. Timing: from 11:20 to 9:50. His plan is the outline's 540 seconds. The first rehearsal runs 11:20, which is 680 seconds, 140 seconds over.
| Slide | Plan (s) | Rehearsal 1 (s) | After cuts (s) | Final (s) |
|---|---|---|---|---|
| 1 Title and question | 20 | 20 | 20 | 20 |
| 2 Definition | 50 | 60 | 60 | 60 |
| 3 Types | 60 | 60 | 60 | 60 |
| 4 How generative AI works | 60 | 90 | 60 | 60 |
| 5 Daily life | 50 | 50 | 50 | 50 |
| 6 Business | 60 | 75 | 75 | 60 |
| 7 Numbers | 60 | 75 | 75 | 60 |
| 8 What goes wrong | 60 | 70 | 70 | 70 |
| 9 Rules | 50 | 60 | 60 | 60 |
| 10 What's next | 40 | 70 | 40 | 40 |
| 11 Conclusion | 30 | 50 | 50 | 50 |
| 12 Sources | 0 | 0 | 0 | 0 |
| Total | 540 (9:00) | 680 (11:20) | 620 (10:20) | 590 (9:50) |
The biggest overruns were slide 4 (90 − 60 = 30 seconds: too much on training) and slide 10 (70 − 40 = 30 seconds: three trends instead of two). Cutting both back saves 60 seconds: 680 − 60 = 620 seconds, 10:20. Trimming 15 seconds of pauses from each of slides 6 and 7 gives 620 − 30 = 590 seconds, 9:50, with 10 seconds to spare. Check the final column: 20 + 60 + 60 + 60 + 50 + 60 + 60 + 70 + 60 + 40 + 50 + 0 = 590.
Arjun's 10-slide version of this deck, with the survey, the incidents pair and the timing log, was built in NOXI from his notes.
Here are two slides from this article's worked example, as NOXI built them, unedited.
Pass/fail checklist before you present
Before an AI presentation, run these seven pass/fail checks.
- A question on slide 1, answered on slide 11.Pass: the answer is one sentence. Fail: the last slide says "Thank you".
- One definition with its source.Pass: OECD or another named body. Fail: an unsourced definition or none.
- Every number has a source line and a year.Pass: organisation, report and year under each figure. Fail: any bare number or "recently".
- One risk with a number.Pass: for example 362 documented incidents in 2025. Fail: "AI is dangerous" alone.
- One rule with two dates.Pass: a past and a coming date from an official page. Fail: "governments are regulating it".
- Timed within the slot.Pass: the last rehearsal ends with at least 10 seconds to spare. Fail: any run over the limit.
- Images of real uses.Pass: phones, maps, scans, inboxes. Fail: a glowing robot on the title slide.
3 AI prompts for an artificial intelligence presentation
AI can draft an AI presentation in minutes if the prompt fixes the question, the slide count, the field and the rule that every number carries a source line and a year. These prompts work in NOXI and in most AI presentation tools.
Create a 12-slide presentation on artificial intelligence for [my second-year BBA class], 10 minutes, answering [Should a small business use AI in 2027?]. Slides: 1 question; 2 the OECD definition of an AI system; 3 types as nested circles (AI, machine learning, deep learning, generative AI); 4 how generative AI works in three steps; 5 three daily-life examples; 6 AI in [business]: one use, one benefit, one limit; 7 three figures from the Stanford HAI 2026 AI Index with source lines and years; 8 what AI gets wrong, with documented incidents (233 in 2024, 362 in 2025); 9 EU AI Act dates and the NIST AI RMF; 10 two dated trends; 11 the answer; 12 sources. No robot images.
Build a 15-minute, 12-slide seminar on [AI in healthcare] for [third-year nursing students], opening with [Would you let AI read your X-ray?]. Slides 6–8: one real use, the evidence with a named official source and year, and the limit (robot-assisted surgery is high-risk under the EU AI Act). Show the four EU AI Act risk levels as a table. A source line with organisation and year under every number; end with a one-sentence answer and sources.
Turn this deck into an 8-slide, 5-minute version for [class 8 students]. Keep the question, a simple definition ("a computer program that learns from lots of examples, then makes a guess about something new"), one daily-life example per slide, one slide on what AI gets wrong and one rule slide. Remove investment figures and the NIST slide; replace robot images with a phone, a game or a map; keep every source line.
How to build an AI presentation in NOXI
NOXI builds the 12-slide deck from your prompt. NOXI is free, with no card and no watermark, and a full deck uses about 25 of the 1,000 monthly credits each account gets.
Step 1: Describe the presentation
Choose Create from scratch and paste Prompt 1 with your question, audience, field and time limit. Because you start from a prompt, NOXI researches the topic on the web first, and slides with figures carry a source line linking to where each number came from.
Step 2: Review the outline first
Choose Plan mode to review the outline before any slide is drawn. Check for 12 slides, the question on slide 1 and its answer on slide 11; then generate and watch each slide being built.
Step 3: Check every number against its source
Open the source line on each figure slide and confirm the number, its year and its wording on the linked page. Ask the AI to add the data year wherever one is missing.
Step 4: Edit by chat
Ask the AI in plain English to change one slide or the whole deck, or click any element and say what to change: "turn slide 9 into a four-row table of the EU AI Act risk levels".
Step 5: Adapt it to your audience
Use Prompt 3 for the 8-slide class 6–10 version, or ask for the 6-slide work briefing. Edit text right on the slide and swap in your own photographs.
Step 6: Present or download
Present full-screen straight from NOXI, share a view-only link with your teacher, or download an editable PowerPoint (.pptx) with real text boxes and embedded fonts, or a PDF (downloads need a free account).
For a comparison of AI tools students use for class decks, see our ranking of the best AI presentation makers for students.
Frequently asked questions
- What is artificial intelligence in simple words?
- Artificial intelligence is a computer program that learns patterns from many examples and uses them to make a guess about something new: a prediction, text or an image, a recommendation or a decision. The OECD defines an AI system as a machine-based system that infers from its input how to generate such outputs.
- What should a presentation on AI include?
- A presentation on AI should include one sourced definition, the main types, three everyday examples, one field in depth, three numbers with source lines and years, one risk with a number, one law or framework with dates, and a conclusion that answers the question you asked on the first slide. Finish with a sources slide.
- How many slides should an AI presentation have?
- For a 10-minute AI presentation, use 12 slides planned at 540 seconds, leaving one minute spare. For 5 minutes with younger students, use 8 slides; for a 5-minute work briefing on generative AI, use 6. If an assignment asks for exactly 10 slides, merge the types and how-it-works slides and drop the final sources slide.
- What are the types of AI?
- By capability, AI is narrow (built for specific tasks, like the AI in today's products) or general (a hypothetical system as capable as a person across tasks). By method, machine learning learns from data, deep learning is machine learning with many-layered neural networks, and generative AI is deep learning that creates new content.
- What is generative AI?
- Generative AI is artificial intelligence that creates new text, images, audio, video or code in response to a prompt. It is trained on large datasets, learns their patterns, and builds an answer by predicting what is most likely to come next, piece by piece. That is why its answers can sound fluent and still be wrong.
- What are good AI topics for a presentation?
- Good AI topics are questions about one field with a source you can cite: AI in healthcare, education or business, bias in AI hiring, deepfakes, AI and privacy, AI and jobs, AI agents or the EU AI Act. Younger students do well with how a chatbot writes and AI in daily life.
- How is AI used in daily life?
- People use AI every day, often without noticing: email spam filters, video and music recommendations, map apps predicting journey times, phone cameras adjusting photos, website translation and chatbots answering questions. The EU AI Act names spam filters and AI in video games as minimal-risk examples, and chatbots as a transparency-risk example.
- What are the risks of AI?
- The main risks of AI are confident errors, bias in decisions about people, privacy problems with the data it uses, misuse such as deepfakes, and harms recorded as AI incidents. The 2026 AI Index counts 362 documented AI incidents in 2025, up from 233 in 2024, a rise of 55.4%.
- What is the EU AI Act?
- The EU AI Act is the European Union's law on artificial intelligence. It entered into force on 1 August 2024 and became applicable on 2 August 2026, with some exceptions. It sorts AI into four risk levels: unacceptable (banned), high (strict obligations), transparency (disclosure duties) and minimal (no specific rules).
- Can AI make a presentation about AI?
- Yes. AI presentation tools can draft a full deck about artificial intelligence from a prompt that fixes the question, the slide count, the field and the rule that every number needs a source line and a year. You still check each number against its source, choose examples your audience knows and rehearse to the time limit.
Sources & methodology
How this guide was made: the definitions come from the OECD, AI figures from the top takeaways of Stanford HAI's 2026 AI Index Report, AI Act dates and risk levels from the European Commission's AI Act page (last updated 19 October 2026), and framework details from NIST. The outline, seconds per slide, versions, topics and checklist are editorial recommendations. Arjun Mehta, his survey and his timings are fictional; every calculation was checked twice. All sources were checked on 21 October 2026.
- [1] Organisational adoption of 88%; 4 in 5 university students now use generative AI; generative AI at 53% population adoption within three years, faster than the PC or the internet; 362 documented AI incidents, up from 233 in 2024 (a record 362 in 2025, per the chart caption); US private AI investment of $285.9 billion in 2025, more than 23 times China's $12.4 billion; over 80% of US high school and college students use AI for school tasks, half of middle and high schools have AI policies, 6% of teachers say they are clear; a top model reads analogue clocks correctly 50.1% of the time; OSWorld agent success from 12% to about 66%; 5,427 US data centres; 73% of experts vs 23% of the public; trust in government to regulate AI 31% (US) and 91% (Singapore) — Stanford HAI, The 2026 AI Index Report: top takeaways.
- [2] AI Act in force 1 August 2024; prohibitions and AI literacy from 2 February 2025; governance and general-purpose AI obligations from 2 August 2025; applicable 2 August 2026 with some exceptions; transparency rules from August 2026; prohibition 9 (non-consensual sexual content and CSAM) from December 2026, introduced as part of the AI Omnibus; high-risk rules for Annex III uses extended to 2 December 2027 and for Annex I products until 2 August 2028; AI Omnibus adopted 19 November 2025, political agreement 7 May 2026, in force 27 July 2026; four risk levels, nine prohibited practices and the examples cited; page last updated 19 October 2026 — European Commission, AI Act (Shaping Europe's digital future).
- [3] AI RMF 1.0 released 26 January 2023, intended for voluntary use; Generative AI Profile NIST-AI-600-1 released 26 July 2024; concept note for a critical-infrastructure profile on 7 April 2026; AI RMF 1.0 being revised as part of the White House AI Action Plan — NIST, AI Risk Management Framework.
- [4] The AI RMF Core is composed of four functions: govern, map, measure and manage, with their descriptions — NIST Trustworthy and Responsible AI Resource Center, AI RMF Core.
- [5] The updated OECD definition of an AI system and its alignment with policy and regulatory processes internationally, including in the European Union and Japan (29 November 2023) — OECD.AI, Updates to the OECD's definition of an AI system explained.
- [6] Definitions of an AI incident and an AI hazard and the harms they cover (17 May 2024) — OECD.AI, Name it to tame it: Defining AI incidents and hazards.
Build your artificial intelligence presentation in NOXI, free: describe your question and field, review the 12-slide outline in Plan mode, check every source line, then present full-screen or download an editable PowerPoint.
