Joe Sack | JoeAI | blog
Could you explain AI to your mum?
Not “AI is a paradigm shift leveraging neural network architectures to optimise workflows.” Actually explain it. In words a normal person would use, over a cup of tea, without anyone’s eyes glazing over.
I tried recently. My mum had questions. Lots of questions. She is starting to send me videos about Chat-GPT she shared me a conersation that she was having with Chat-GPT about the foxes she has living in her garden. So luckily my mum is asking good questions about AI and how the LLMs work. The kind most articles skip because the people writing them have already forgotten what it’s like not to know.
So I wrote down every question she asked me. And I answered them the way I answered her. No assumed knowledge. No showing off. Just plain English.
If you’ve been hearing about AI everywhere but still don’t really get what any of it means, this is for you.
I’m Joe. I run JoeAI, a small AI consultancy based in Crouch End, North London. I help small businesses use AI without the overwhelm. This is where that starts.
What actually is AI? and is it different to Google
Artificial intelligence. Software that can do things we used to think only humans could do. Recognising faces in photos. Translating languages. Having a conversation. Writing an email. Summarising a long document into three paragraphs.
It is not a robot. It is not alive. It is not thinking the way you and I think. It is software that has been trained on enormous amounts of text, images or data, and has learned patterns from all of it. When you ask it a question, it is not looking up the answer in a filing cabinet. It is predicting what the most useful response would be, based on everything it has seen before.
Think of it like this. If you read every cookbook ever written, you would probably be able to come up with a pretty good recipe for a chocolate cake, even if nobody asked you to memorise one. You would just know what usually goes together. That is roughly what AI is doing. Pattern recognition on a massive scale.
When did it start? I thought that AI was a new thing.
The idea has been around since the 1950s. A mathematician called Alan Turing asked the question “can machines think?” back in 1950. Researchers have been working on it in universities ever since.
But for decades it was mostly academic. Slow. Limited. Not something ordinary people would ever use.
What changed was ChatGPT. It launched in November 2022 and reached 100 million users in two months. That is the fastest any technology product has ever grown. Suddenly anyone with a web browser could have a conversation with an AI and get a useful answer. That is when the rest of the world started paying attention. Our brother in law was playing around with his phone and decided to write a long poem about his aunty for her birthday dinner. The table was moved by the heartfelt message but it turns out Chat-GPT had built the sublim birthday message through a couple of facts about her life. He then also started to make songs and funny images about different family members and is a constant source of amusement (annoyance?)
So the technology is not new. But the version of it that you and I can actually use? That is brand new. We are still in the early days.
What’s ChatGPT then and why is it thought of as the Ipod of AI?
ChatGPT is a product made by a company called OpenAI. It is the most well known AI tool in the world right now. They are currently one of the biggest companies in the world at the moment and a lot of that is being the Go-To Ai model that everyone thinks about.
You type a question or a request into a chat box. It types back an answer. You can ask it to explain something, write a letter, summarise a document, help you plan a holiday, draft an email, or work through a problem with you.
It is not the only AI tool. It is just the one that got famous first.
Who are OpenAI and how are they related to Chat-GPT
OpenAI is the company behind ChatGPT. They were founded in 2015 by a group that included Sam Altman and Elon Musk, originally as a nonprofit research lab. The idea was to build AI safely for the benefit of everyone.
Since then it has become a commercial company worth hundreds of billions of dollars. Musk left and has started his own AI company called xAI. The “open” part of OpenAI has become a bit of a debate, because the technology is not actually open for anyone to inspect or use freely. But the product works well and hundreds of millions of people use it.
You keep talking about Anthropic and Claude. Who is Claude, Do you know him? Are they the same person ?
Anthropic is the company that makes Claude, which is the AI I use most in my own work.
It was founded in 2021 by Dario and Daniela Amodei, who are brother and sister. They both previously worked at OpenAI but left because they wanted to take a different approach to AI safety. Their whole thing is building AI that is powerful but also honest, harmless and transparent about what it does and does not know.
As of 2026, Anthropic is valued at around $380 billion and their AI models are used by millions of people and businesses worldwide. They are based in San Francisco.
What’s Claude and what kind of things can it do?
Claude is Anthropic’s AI assistant. It is the equivalent of ChatGPT but made by a different company with a different philosophy.
You use it the same way. You type something in, it responds. You can ask it questions, have it write things for you, analyse documents, brainstorm ideas, check your work.
I personally find Claude better for writing, reasoning and longer pieces of work. Claude is also one of the best on the markets for coding. You will find that models are good and on thing and not so good at another. They have different personalities. ChatGPT has a bigger ecosystem of plugins and image generation but it also agrees with almost everything you say. Claude has different tools for different jobs but it more likely to push back on you when it comes to having a really out there Idea. But both are genuinely excellent they just need to have different guard rails to keep them to task.
Claude is the AI I use daily in my consultancy. It is also the one generating this draft for me to edit (yes, I practice what I preach).
Are all the AI tools at war with each other?
Sort of. But not in the way you might think.
OpenAI, Anthropic, Google and a handful of others are all competing to build the most capable AI. There is serious money involved. Billions of dollars in investment. The talent moves between companies. The press loves a rivalry as well so this sort of Soap opera is the best of both worlds. It is real it is fast moving and like eastenders you need to keep your finger on the pulse with it otherwise you will loose track.
But it is more like a technology race than a war. Each company releases new models, the others respond, everyone improves. The competition is genuinely making the products better for people like you and me.
It is similar to how Apple and Samsung compete on phones. They are rivals. They push each other. But you benefit from the competition whether you pick one or the other.
The real tension is not between the companies. It is about how fast to move versus how carefully. Anthropic tends to be more cautious about safety. OpenAI tends to move faster. Google has the most data and the most existing users. Each has a different bet on what matters most.
What’s Opus and Sonnet? They sound like perfume.
Ha. Fair enough.
Opus and Sonnet are the names Anthropic gives to different versions of Claude. Think of them like different sized engines in the same car brand.
Opus is the biggest and most powerful. It is the best at complex reasoning, difficult problems and long, detailed work. It is also the most expensive to run. Since i started JoeAI we are now moving from Opus 4.6 to being on Opus 4.8. Claude is constantly building new tools and features to differentiate themselves from the other competitors.
Sonnet is the middle tier. Faster, cheaper, still very capable. For most everyday tasks, Sonnet is more than enough.
There is also Haiku, which is the smallest and fastest. Great for quick, simple jobs where speed matters more than depth.
If you are using Claude through the website or app as a regular person, you do not need to worry much about this. The system generally picks the right one for you. But it is worth knowing the names exist so you do not feel lost when someone mentions them.
What’s a token and how many of them do i have?
This is one of those words that sounds technical but is actually simple.
A token is a chunk of text. Roughly three quarters of a word. The word “basketball” is two tokens. The word “the” is one token.
AI tools measure everything in tokens. How much text you send in (input tokens) and how much text comes back (output tokens). The more tokens involved, the more it costs to run.
For most people using ChatGPT or Claude through the website, you never see tokens. You just type and get an answer. But if you are a business using the technology behind the scenes through the API (more on that in a moment), tokens are how the bill is calculated.
Think of tokens like units on a mobile phone plan. You do not need to know exactly how many you are using every time you send a text. But it helps to understand that the meter is running.
Top Tip: When you have a conversation with claude your chat builds up a context layer of tokens if the chat gets too big than basically the ammount of tokens you use goes up dramatically. It is better to start a new chat than to carry on.
What’s a prompt?
A prompt is just what you type into the AI. Your question. Your instruction. Your request.
“Write me a thank you email for my client” is a prompt. “Explain photosynthesis to a ten year old” is a prompt. “Summarise this document” is a prompt.
The better your prompt, the better the answer. That is why people talk about “prompt engineering” which just means getting better at asking clear questions. It is not magic. It is the same skill as writing a good brief for a designer or giving clear instructions to someone new at work.
What’s a large language model?
This is the technology behind ChatGPT, Claude and most of the AI tools you hear about.
A large language model, or LLM, is a piece of software that has been trained on a vast amount of text. Books, articles, websites, conversations. Billions of pages of it. Through that training, it has learned how language works. How sentences are structured. What words tend to follow other words. What a good answer to a question usually looks like.
It does not understand the world the way you do. It does not have experiences or feelings. But it is extraordinarily good at producing text that is useful, relevant and often surprisingly insightful.
“Large” refers to the size of the model. More parameters (the internal settings the model has learned) generally means more capable. The biggest models have hundreds of billions of parameters.
What’s an API?
API stands for Application Programming Interface. I know that does not help.
Here is the simple version. When you use ChatGPT or Claude through the website, you are using the product. When a developer connects their own software to the AI so it runs in the background, they are using the API.
The API is like the engine without the car body. Same power, but you build your own vehicle around it.
For example, a therapy practice might use the Claude API to power an automated client intake form on their own website. The client never sees Claude. They just see a smart form that asks good follow up questions. Behind the scenes, the AI is doing the work.
Most small business owners will never touch an API directly. But it is useful to know the word exists because it comes up a lot, and because it is the difference between using AI as a consumer and using it as a building block for your business.
What’s Perplexity? Someone told me to try it.
Perplexity is an AI powered search engine. Instead of giving you a list of links like Google does, it reads multiple web pages for you, pulls out the relevant information and gives you a written answer with sources cited.
Think of it as having a research assistant who reads everything and gives you a summary with footnotes.
It launched in 2022 and has grown quickly. As of 2026 it has over 45 million monthly users and handles more than a billion searches a month. It is free to use, with a paid tier that gives you access to more powerful AI models and deeper research features.
The key difference from ChatGPT or Claude: Perplexity is connected to the live web. It searches in real time. ChatGPT and Claude have knowledge cutoffs and need to be specifically told to search the web. Perplexity does it by default.
If you want to research a topic and get a clear answer with sources you can check, Perplexity is the best starting point.
What about Google? Are they doing AI too?
Very much so. Google has its own AI called Gemini (it used to be called Bard). They also own DeepMind, one of the most advanced AI research labs in the world.
Google’s advantage is data. They have Search, YouTube, Gmail, Maps, Android. More information about how people use the internet than anyone else alive. They are building AI into all of it.
The reason Google sometimes feels behind is not because their technology is worse. It is because they have more to lose. Their entire business is built on search ads. AI that just gives you the answer threatens the model where you click on ten links and see advertisements alongside them. So they are moving carefully.
But make no mistake. Google is one of the biggest players in AI and will be for a long time.
Is AI going to take my job?
This is the big one.
The honest answer is: probably not in the way you are imagining. But it will almost certainly change your job.
AI is very good at tasks. Summarising documents. Drafting emails. Processing data. Answering routine questions. Scheduling. Generating first drafts.
AI is not good at relationships. Understanding a client’s unspoken concern. Reading a room. Making a judgement call that requires experience and context. Leading a team. Building trust with someone face to face.
The pattern we are seeing is that AI handles the repetitive, time consuming parts of a job, freeing the human to do the parts that actually require a human. Therapists who use AI for session notes spend less time on paperwork and more time with clients. Accountants who use AI for data entry spend more time advising.
The people most at risk are the ones who refuse to learn how any of it works. Not because AI will replace them directly, but because someone who does the same job and uses AI alongside it will be faster, cheaper and often better.
The best thing you can do is start learning. One tool. One task. That is enough for now.
Is it safe? Should I be worried?
You should be informed, not worried.
AI tools like ChatGPT and Claude are safe to use for everyday tasks. But they have real limitations you need to know about.
They can be wrong. Confidently wrong. They will give you an answer that sounds completely authoritative and is completely made up. This is sometimes called a hallucination. Always check anything important.
They are not private by default. If you type sensitive business information or personal data into a free AI tool, that data may be used to train future versions of the model. Use business grade versions with proper data handling if you are working with anything confidential.
They do not know what they do not know. An AI tool will not tell you it is missing context. It will just give you the best answer it can with whatever you gave it. That is how I ended up wasting twenty minutes on a problem that had a ten second solution sitting right there on my own website.
Used well, with a human checking the output, AI is an incredibly useful tool. Used blindly, it will waste your time and might lead you somewhere wrong.
That is not a reason to avoid it. It is a reason to learn how to use it properly.
Where should I start with AI?
One tool. One task. That is it.
If you have never used AI before, go to claude.ai or chatgpt.com. They both have free versions. Type in a question about something you are working on this week. See what comes back. Edit it. Try again with a better question.
Do not try to automate your entire business. Do not sign up for fifteen tools. Do not read a hundred articles about which model is best.
Just start a conversation and see if it saves you time on one thing. That is the only test that matters.
If you would like some help working out which tool fits your situation, I run free AI workshops for North London professionals and offer one to one consultations for anyone who would rather talk it through with a real person.
Want to talk it through?
I run free AI workshops for North London professionals. If you would rather talk one to one, a £75 introductory consultation is the place to start. No hard sell. Just a practical conversation about what would actually help.
Get in touch at joe@joeai.co.uk or visit joeai.co.uk/contact
Joe Sack is an AI consultant based in Crouch End, North London. He helps small businesses and independent professionals use AI without the overwhelm. Over ten years in digital marketing working with brands including Volkswagen and Unilever, now focused on making that same thinking accessible to the people who need it most.
