📫 Subscribe | 🗣️ Hire us to speak | 🤝 Sponsor this newsletter | 🎓 Events

Welcome to the Friday edition of our newsletter. We spend Fridays going deeper into tools and trends related to generative AI (and Tuesdays sharing news updates). This week, Professor Snider is sharing what AI knowledge he feels his students need to have before entering the real world.

We’ll be greeting our incoming first-year students at Drake University in less than a week. So I have been thinking about what they will need to know before they graduate ... in 2030!

One of my jobs is to figure out what AI knowledge students need before they walk across the stage at graduation - whether next year or in 2030. I also serve on a committee thinking about this beyond just the School of Journalism where I teach. Every student - accounting majors, biology majors, education majors, all of them - need a certain level of AI knowledge.

Here’s my working answer: a baseline of AI knowledge that every graduate (and honestly, every working professional) should have. Half of this is about using these tools well. The other half is about surviving a world full of what the tools produce. See what you think.

You don’t need to understand the math behind AI models. But you should know this: generative AI tools like ChatGPT, Claude and Gemini are prediction machines. They were trained on massive amounts of text, and they generate responses by predicting the next most likely word, over and over, really fast.

That’s it. No database of facts. No understanding the way you and I understand things.

Why does this matter? Because once you know it’s predicting rather than looking things up, everything else about AI starts to make sense - including why it gets things wrong.

AI tools make things up. Confidently. They’ll invent court cases, fake citations, wrong statistics and quotes nobody ever said.

This isn’t a glitch that will get patched next month. It’s baked into how prediction machines work. The models are getting better, but the risk never goes to zero.

The baseline skill: never publish, submit or forward an AI-generated fact you haven’t verified yourself. Ask the AI for its sources, then actually check them. Ask it what things it assumed, and then check those.

These tools learned from the internet - and the internet has opinions. AI models reflect the biases in their training data, from whose perspectives get centered to which names show up in example resumes.

A graduate should be able to ask: Who might be left out of this response? Whose viewpoint is this? Users should question the output instead of simply accepting it.

The difference between a useless AI response and a great one is usually the prompt. The basics every graduate should know:

At Innovation Profs we teach this as the BRIEF method (Background, Role, Instructions, Examples, Format). Whatever framework you use, the point is the same: vague in, vague out.

There are hundreds of AI tools and new ones every week. Here’s the shortcut. Almost everything falls into a few buckets.

The general-purpose chatbots - ChatGPT, Claude, Gemini, Copilot - do roughly the same core work: writing, summarizing, analyzing, brainstorming, answering questions. Then there are the specialists: image and video generators, meeting notetakers, research tools, coding assistants.

For most people at most jobs, the right general-purpose tool is the one your employer already pays for. Not because it’s the best. Because it’s the one with the data protections (see No. 6).

Once you are in a chatbot, you likely need to choose which model to use.

Open ChatGPT or Claude or Gemini and there’s a dropdown with names that mean nothing to most people. The names change every few months. Don’t memorize them. Learn the three flavors instead:

Fast (the default): Quick answers, first drafts, summaries, everyday questions. This is 80% of what you need.

Thinking or reasoning: Slower. Works through the problem step by step before answering. Use it for analysis, math, multi-step planning, or anything where being wrong is expensive.

Light or mini: Cheap and fast for simple, high-volume tasks. Often what you get bumped to when you hit a usage limit.

Stay on fast, step up to thinking when the task has multiple steps or a real cost of being wrong. Don’t burn a reasoning model on a social media post.

AI isn’t a separate tool anymore. It’s inside the tools people already use - Microsoft 365, Google Workspace, Adobe products, Canva, your email, your CRM.

Graduates should walk into a job knowing the common use cases: drafting and editing, summarizing long documents, brainstorming, data analysis, meeting notes, first-pass research. And they should know which tasks AI handles well (first drafts, summaries) versus where it struggles (final drafts, anything requiring judgment or current facts).

A baseline includes knowing what NOT to put into an AI tool: confidential company information, client data, anything covered by privacy laws, unpublished research.

If you wouldn’t post it publicly, don’t paste it into a free AI chatbot. Know whether your employer has an approved tool with data protections - and use that one.

When should you tell someone AI helped with your work? We wrote about this a couple weeks ago.

The short version: it depends on the context, but the instinct to ask the question is the baseline. A graduate should understand that passing off AI work as fully their own can damage trust - with a professor, a boss or an audience. And they should know that “the AI said so” is never a citation. Cite the original source, not the tool.

Everything above is about being a good user of AI. This one is about being a good consumer of everything AI produces - which is now a big chunk of what shows up in your feed, your inbox and your search results.

Stop trying to spot the fakes. For a couple of years the advice was to count the fingers and look for weird text in the background. That advice is dead. AI images and video are good enough now that eyeballing it doesn’t work, and neither do most “AI detector” tools - they produce false positives constantly.

Ask where it came from instead. The question shifts from “does this look real?” to “who posted this, and where did they get it?” That means checking the original account, doing a reverse image search, looking for the same photo or clip in a credible outlet, and being suspicious of anything with no traceable origin. Provenance beats appearance. It’s the same verification instinct journalists have used for decades - it just matters for everyone now.

Know about the liar’s dividend. The bigger danger isn’t that you’ll believe a fake video. It’s that AI gives everyone a reason to dismiss real evidence. Caught on camera? Just call it AI. When anything could be fake, everything becomes deniable - and that’s corrosive in a way that’s harder to fix than any single deepfake.

Understand the AI answer layer. Google AI Overviews, ChatGPT, Perplexity - more and more people get information as a synthesized answer with the sources buried or stripped out. That’s a different information diet than a list of links. Graduates should know that the summary is a prediction of the answer, not the answer, and they should still click through to the source that matters.

Recognize AI slop and engagement farming. There are entire sites, Facebook pages and YouTube channels generating AI content at volume purely for clicks and ad money. Knowing that this exists - and that a lot of it is designed to make you angry enough to share - is a baseline defense.

Nobody can predict AI’s future with certainty. But graduates should understand the direction of travel: AI agents that complete multi-step tasks, AI built into every piece of software, and job descriptions that increasingly assume AI fluency (we’ll discuss recent developments in our upcoming AI Update).

They should also know what AGI means, because they are going to hear it constantly.

AGI stands for artificial general intelligence - roughly, an AI that can handle most things a human can, across subjects, without being trained separately for each one. It’s what every major AI lab says it’s building. It’s the reason for the hundreds of billions of dollars going into data centers.

The practical takeaway: the skill isn’t mastering one tool. Tools change monthly. The skill is staying curious, experimenting regularly and being able to evaluate a new AI tool when it lands on your desk.

If I had to compress all of this into one sentence, it’s this: know what AI is good at, know where it fails, and never outsource your judgment to it - whether you’re the one prompting or the one scrolling.

What would you add to this list? If you’re hiring right now, what AI knowledge do you wish new graduates showed up with? Hit reply - I would love to hear what you’re thinking.