5 AI Terms Every Beginner Has To Know (AI Dev Explains)
March 17, 2025
This video offers a practical primer on a handful of core AI terms, with concrete prompts and recommendations you can use today to get better results from AI tools.
Key concepts: LLMs, tokens, and context#
- Large Language Model (LLM)
- A model trained on a massive amount of text to guess the next word in a sequence.
- It’s excellent at predicting how people talk and generating coherent text.
- Tokens
- The unit a model uses to count and process text (not just letters, but meaningful chunks of text).
- Think of tokens as the “how the computer counts words” mechanism rather than just individual characters.
- Context window
- How much of the conversation or prompt the model can “see” at once.
- If you push past this limit, the model can start to misbehave or hallucinate because it has no more context to rely on.
- Analogy: trying to fit a huge bag into a small overhead bin at the airport; once it can’t fit, the process starts to break down.
Hallucination: what it is and why it happens#
- Hallucination
- The model makes stuff up with high confidence.
- It happens when the model doesn’t have enough context or is overextended beyond its window.
- Practical takeaway
- Keep prompts within the model’s context window.
- Treat outputs as best-effort guesses and verify critical details.
Prompting: levels and how to get better results#
- The word prompt has become central in AI work, but the real power is in how you structure it.
- Prompt levels (a practical mental model)
- JV (junior varsity) prompt: a simple request like “Write me a sales email.”
- Quick, but likely missing specifics, context, and tone.
- Varsity prompt: more guidance and role assignment.
- Example: “You are an expert sales writing assistant. I have a draft; improve it with a focus on benefits, clarity, and a strong CTA. Here are guidelines and examples.”
- Key ideas to apply
- Define the role: tell the model what it is (e.g., “You are an expert editor”).
- Provide context: what the output is for, target audience, benefits vs features.
- Give examples: show samples that you like.
- Focus on outcomes: emphasize benefits and customer pain relief.
- Practical note
- There’s a whole continuum of prompting; a little more upfront work yields much better results.
Models and where to use what#
- Types of models
- Chat models: the back-and-forth conversational format.
- Chain-of-Thought (reasoning) models: think through the problem, reason step by step, possibly with a longer deliberation before answering.
- Popular players (as discussed)
- Claude (Anthropic)
- Strong for writing tasks; different variants exist (e.g., Claude 3.5 Sunet and related versions).
- ChatGPT (OpenAI)
- Various models with different “O” configurations (e.g., 03 mini for a blend of speed and reasoning, 4o for more capable reasoning, 01 for longer consideration).
- Has many models, but the practical UX and writing quality may vary; still a relevant option to explore.
- Budget and practicality
- You’ll pay for access to higher-quality models; the cost is usually modest relative to the value (think a few dollars per month).
- If you’re primarily writing, Claude is highlighted as a standout option in the video’s view.
Beyond text: multimodal and content transformation#
- AI can handle more than writing
- Image generation, video generation, and even audio/text transformations.
- You can transform content types (blog to video, script to podcast, etc.) and mix modalities (text-to-speech, speech-to-text, etc.).
- Examples mentioned
- Video and image workflows, but also “content pipelines” where one form feeds another (e.g., a script converts into a YouTube video concept).
- Emerging multimodal tools and continued AI arms race (new models launching, ongoing improvements).
Practical takeaways you can use today#
- Learn these basics first:
- LLM, tokens, context window, hallucination, prompting, and model types.
- Start prompting with intent:
- Define roles clearly, provide context, and include examples to guide outputs.
- Pick the right tool for the job:
- For writing: experiment with Claude or a capable ChatGPT option; choose the variant that fits your needs (speed vs depth).
- For reasoning-heavy tasks: consider models that emphasize chain-of-thought or longer deliberation.
- Use AI to accelerate workflows, not just replace them:
- Convert between content types, brainstorm ideas, draft materials, and then manually refine the outputs you get.
Quick ideas to experiment with#
- Draft a sales email using a JV prompt; then rewrite it with a Varsity prompt for tone, benefits, and a strong CTA.
- Take a piece of content (e.g., a blog post or script) and generate a YouTube video outline or a LinkedIn post series.
- Use prompts to convert a photo or design concept into descriptive copy or a script.
Links#
- Claude (Anthropic) — Claude 3.5 Sonnet and related variants
- ChatGPT / OpenAI — GPT-3.5, GPT-4, and related model options
- Google AI models — various available models
- Grok — Elon Musk's xAI model launch and discussion
- Goldman Sachs AI Report — 300 million jobs could be affected by AI automation
If you found this primer helpful, note practical prompts you’ll try first and share your favorite use cases in the comments.
Transcript
[Music] check this out there's a literal fire hose of information on AI and even as a full-time developer I have trouble keeping up you know that thing mors law it's like hey the speed to which the internet's going to speed up is going to get crazier and crazier and crazier well just imagine that but like 10x it like it is nuts and if you're a developer and you're on X you're in this little bubble and you think that all this stuff is just normal and if you don't know all the terms that you're outside of it guess what like 99.9% of planet Earth has no idea about all this crazy stuff that's happening and it doesn't even matter that they don't know all the terms it just matters like how will this help them because Goldman Sachs is talking about how 300 million jobs are going to be displaced by 2030 that's kind of crazy so it's definitely worth knowing I'm not saying that as like a scare tactic I'm just it's a stat it's kind of nuts but we're going to show these terms on the screen so let's start with llm you maybe heard of it you know AI cuz you clicked on this video but what does that mean well it means large language model and what that is is basically a computer gobbled up a bunch of the internet like we're talking a lot of Internet you think when you're staying up late you watch a little bit of the internet that you're consuming a lot of Internet well take that and multiply it by like a trillion and then you end up with a lot of internet and that's what llms do that's what they've done when they're training they're basically getting really really good at learning how people talk how the text lines up word after word after word because when it goes to respond to something it is guessing what the next word is it's really good at guessing though so that's what an llm is and now we're going to talk about what a token is you probably either think of being at a casino or maybe like cryp do scammy stuff well guess what my fiance was very confused today because we're talking to this new AI model we're testing it out and I was asking about tokens she's like what you have to pay for this in tokens no you don't pay for it in tokens but a token is kind of like how the computer thinks about words it's their way of counting words when I think of the letter c a or the letter c a for cat there's three characters in there that's kind of how we denote things but an llm thinks of them in token the easiest way that you could kind of think of this is if you had a sentence like the quick brown fox jumps over the lazy dog well that's a really popular sentence in the English language so if you're going to guess what that last word is it's probably going to be dog right like how often are you going to say that the quick brown fox jumps over the A5 wagu or like rubber duck or something probably not going to say that probably say dog so it assigns a best guess as to what the next word was the more the thing trains on the more information SL interet it's gobbled up the better it's going to be at guessing the word dog and some llms well all of them they have what's called a context window what does that mean how much information can be held in a conversation with one of these chat Bots before it starts to either stop working or lose its mind where it's hallucinating that's another term hallucinating it means it's just like making stuff up but it says it like so confidently that you're like maybe this thing's right you could be like hey write me a sales email about XYZ or give me a statistics about this this and that and it'll come back and just like none of it's right that's hallucination but that happens when the context window aka the amount of stuff that you can fit into it just it goes beyond it there's no more capacity it's like if you went on a flight you got this massive golf bag and you're trying to put it in like the little tiny thing and you're like I can't pay for this bag it fits in no it doesn't fit in there sir go pay the $35 to American Airlines that's kind of like what a context window is and so it's kind of how all this stuff like works together I know we jumped around in the glossery if that's on the screen but the context window means you can fit a lot of stuff in it typically as they get bigger the quality of the output the thing you get back goes down a little bit but at the end of the day these things are getting really good and they're great at guessing the next word whether you're writing a sales email or you're trying to get it to finish your homework or you're trying to get it to code for you to really good at that cuz it's very objective they're really good at guessing now this is probably the juiciest one of them all I no this is not an engagement bait to say like oh stay no the juicy one is prompt cuz when have you heard that word in your entire life before this whole llm AI stuff came along probably never but now it's like the hottest word under the sun because everyone's like oh my gosh if I just get this prompt if I could just like I just need the right prompt I got to get a prompt like they're like freaking out about prompts and for good reason because if you're trying to get some sort of thing back from the magical computer I'm trying to get it it's like your Zool ler you're like it's in the computer you're like trying to pull it out like if you can prompt it the right way then you can get some magic going right but the way I kind of equate this and having spent like three and a half years using these things is you have different levels to this and I'm it's a different video that I have where you can that I'm putting out soon about like levels of prompting but in its most basic sense let's take a sales email and a level zero prompt let's call it like a junior varsity prompt like we're in high school a JV prompt would be write me a sales email okay well it's going to like say something or it might ask you a question about what but just imagine how weird that'd be like if you walked up to someone really smart that like knew your name but you've never talk to them very much like you don't have they have no context about your history your product your service whatever whatever it is you're doing what am I writing this sales email about and then you say oh it's about the car washing business that I have and then it just like starts it could be like if you were on the street and then that person just starts like spewing stuff at you and they're like really confident about it but they don't even know where it is they don't know what the offer is they don't know is it high-end cars mid cars lowend cars do you have a special going on is it open on Sundays it would just miss all that stuff and so the maybe the bump from the JV you know like you did good this year now you're going to Varsity a varsity one might look a little bit more like role play like where you know if you're if you're imagining you're sitting with somebody and they need to be told what to do you'd say hey look like I need help with this I need an edit um this is kind of what we're thinking here yada y y well when you're talking to the computer it's a little different and you need to say hey like you are an expert sales writing assistant and I have the following draft for my company we do X Y and Z and I need you to look at this because our staff editor needs the final that's going to be published in the next hour and it's really important that these 10,000 customers that we're about to reach out to understand exactly what it is that we're doing why we're doing it why it's a benefit to them sell the benefits not the features or the services so the benefits that they're going to get the pain that they're going to be solved or uh relieved of and then here's some examples of some emails that I like now that would be way better cuz we kind of covered these measures where it's like hey like I know what I am like you're you're identifying to the computer what they are like I'm an editor okay and I got to get this out and uh I really need to do a good job cuz like 10,000 I mean I'm kind of role playing what they're thinking but you're basically just pulling the ball forward and you're saying you're doing a little bit more work right like a prompt that you write that's like the level zero JV thing is going to be like the oneliner and you're going to be like oh AI sucks and you're like well no you just don't know how to use it and that's fine totally is fine so models well we have covered that word llm a I ll M needs a Rebrand but the word model that's the m and llm and the model is basically the thing you're talking to so it's the big glob of Internet that got trained on a bunch of computers to figure out how to guess stuff and there're different types the one that you're used to that you probably think of when you clicked on this video is a chat model and a chat model is the back and forth the pingpong game that you play where it's like boom you throw them the ball they're throwing that thing right back might take like 2 seconds Max and they're going to be confident those have been around for a long time um I don't know if I'm like aging myself or anyone that's like over the age of 28 you can think of like smarter child from AOL that thing was kind of like a chat model but not second type of model is a Chain of Thought model now that's a lot they're also called reasoning models thinking models maybe thinking it's probably the easiest to grasp but it's it's like if you threw the ball and then the thing that was catching it in this case the llm or the model it's like going to catch it and like charge it cuz it like really wants to throw that ball back but with like some Force you know and like thought about it it's like I need to throw it over here and the wind is the barometric pressure and uh the temperature today all that stuff it's thinking through and I can throw the jargon at you but it basically takes a little bit of time and it's going to run in a loop so instead of it just like throwing it back with like a oneliner or the first version of the output it will look at what it was going to say and then try again and try again so it's self critiquing itself not to go off on a tangent but there are ones that have deep research now it's going to hold that ball and it's thinking it's it's doing its Chain of Thought thing and then it's going to go search the internet and come back and then bring that in so there's like different flavors of all of this stuff but you just need to know that there's chat models and Chain of Thought models when it comes to the text generation and some of the really popular ones that you should consider using and all these names need a Rebrand I don't know who the high guy was that was naming these but anthropic is the company and they have a model called Claud sounds kind of normal until I say 3.5 Sunnet 1224 25 like what but sunet has 3.7 and 3.5 3.7 came out like two days ago I actually made a video about it cuz I thought I was coming out and I made the video and then it came out and then I made another video got like 40 views huge day for me but it's fantastic at writing it's a little more expensive and the way that they charge by these things is you don't need to worry really you just cuz you're not a developer but sometimes they charge by the token AK the number of things that it's going to spit out for developers but for non you know non- nerds it's like 20 bucks a month and you get access to the best one and it's great at writing so use that one don't use any other seriously don't use anything besides Claude for writing it's just like so good the second is obviously chat GPT it's the one that took the World by storm and they've got a bunch of different models all you need to worry about when you're doing chat or Chain of Thought stuff is does it have an o in it and so if you're paying for these which you should because you literally get access to like a PhD level thing and it's like what the cost of like four Starbucks is or like one 1.5 Chipotle Burritos you get access to the best ones and so their chat ones I would use 03 mini it's like a mix between the chat one and the Chain of Thought one that's holding the ball a little longer so it's going to answer quick but it's going to have a little bit of reasoning if you're just using strictly a chat one then use 4 o that doesn't have an o in front of it but 40 and then 01 is the one that is like going to hold that ball for a while really think about it there's different ways that you talk to him as well but that's for a different video lastly Google that's worth mentioning you know they have gobbled up a lot of internet and I'm surprised that they haven't done more in this space they have a bunch of stuff like don't get me wrong they have a lot of models but they're not very popular and I think it's probably because they don't have a mobile app that's very good but they're known for like giant windows so like they can take all the tokens in and they'll like give them to me but their writing isn't that great and I just haven't used it at all I never use it so if I needed to take the script of every Harry Potter movie and then ask how many times does hermion say vavia levosa that would be a good use case I don't know why I'd ever do that though so I hope that was a good primer on all this nerd stuff because that's what I like doing I like taking all that like I don't know three years worth of using this stuff all the time compress it down cuz I have all these conversation with my friends my fiance where I'm like oh my gosh this is so cool but I need to figure out how to translate it really well and we are benefiting so much from it like the amount that I can get done today on Feb 27 versus 2 years ago one year ago it's astounding but it took a little bit of time to like figure out how to talk to these things it's also worth mentioning that there's different models Beyond those two types that I mentioned like you have video generation which you guys said it generates videos you have image generation which most of them look awful but I figured out a way that you can take any image that you see and and then say Hey how do I make this and then you can use that like blob of text that it's going to give you over and over and over again to maintain a style there's other ones out there you can just jump from one type of content to another so if you have written word and you want to go text to speech you can do that you can go speech to text you can go video to speech I don't know why you want to do that you can go from blog to YouTube you can go podcast to LinkedIn post X post to Instagram real script Instagram real script to a speech like you can just just put the word to in between two different types of content you do sales email sequence to video sales letter you could do project overview to five-page paper like it's nuts there's a ton of different use cases there Croc 3 also just came out that's the one that Elon Musk launched and I had a 30 minute conversation with their doctor which was kind of wild because I don't know like it's it's just one of those things like when the iPhone came out and everyone's like oh my gosh like I had that moment just now and so what's interesting about this arms race that we're in is open AI is supposed to launch a new model in the next couple of days and that's going to be a big deal so I've been calling these as they come out um just cuz I think people need to know about it like I would make this video so that like my mom could watch it and it's kind of my like me it's like is this simple enough to know and um yeah so subscribe to the channel because what else are you going to do what else am I going to say like the video dislike it if you disliked it just kidding don't do that and uh comment below what your favorite AI use case is mine right now is actually just taking a photo of like my fridge being like what can I cook that's decent video idea generation not wrting the whole thing like I just kind of did this off the top because I know this stuff really well but if I was doing something I was like research i' be doing that I think H I'll land on it it's my most used thing is learning software fast so I just am constantly talking to it taking screenshots you can do screenshots in there you can say how do I do this how do I do that all right see you [Music] [Applause] [Music]