What is AI? A plain explanation

You hear the word AI almost every day now. Here is what it actually means, in plain words, and what matters for you as a business owner.

It guesses the next word

When people talk about AI today, they usually mean programs that read and write text. These programs have read an enormous amount of books, websites and other writing. From that, they have learned how words tend to follow each other.

You already know a small version of this from your phone. When you type a message, it suggests the next word. AI does basically the same thing, only much better. It does not look the answer up like a phone book. It guesses which word fits best next, then the one after that, and so on until the answer is finished.

Think of someone who has read a huge amount. They can talk about almost anything and sound convincing. But they do not always remember where they read something, and sometimes they mix up what is true with what merely sounds right.

What it is good at

AI works best when the job is about text:

• summarising a long document in a few lines
• writing a first draft of an email or an advert
• sorting messages, for example whether an email is an order or a complaint
• reading details off an invoice, such as the date and the amount
• finding an answer in a long document when you ask in ordinary words
• translating between languages

It does not get tired, and it does not get bored doing the same task a hundred times a day.

Where it falls short

AI can be wrong. Worse, it sounds just as sure of itself when it is wrong as when it is right. It can make up a number, a name or a source that looks perfectly real but does not exist.

It also knows nothing about your business unless you tell it. It does not know your prices, your opening hours or your customers.

It understands Icelandic but still makes mistakes, for example with word endings or word choice. Its Icelandic can also sound stiff even when the grammar is correct.

And it takes no responsibility. If something goes wrong, a person has to answer for it.

What this means for you

Think of AI as a keen new employee who works fast but does not know the place yet. You let them do the groundwork, and you read it yourself before anything goes out to a customer.

That is why our rule at AkurAI is simple. Where mistakes are costly, such as in bookkeeping or replies to customers, a person always reviews the work before anything is sent or posted. And before we put anything into use, we test it on your own data so you can see how it really performs.

If you want to talk about whether AI could help in your business, drop us a line.