The Walt Disney Company

Nowadays, AI solutions are everywhere around us. Something curious is that most people don’t know that they have been with us since the middle of the twentieth century. We can argue about the beginning of AI and some people would say that it was in the moment that John McCarthy first named it using the term “Artificial Intelligence” in 1956, or if Turing’s machine that was used to decipher the messages encrypted by Germany during the second world war, was in fact the first AI solution even though it has not been named as it. But in any case, AI was born a lot of years ago.

We also use AI daily when we listen to a song or watch a movie just because an algorithm says it will probably like us, or when we talk to our virtual assistants to schedule an alarm, make a phone call or just find the definition of a term on the internet… or when we use a navigation app to find the best route to go home.

We don’t even have a formal and unique definition for AI, but we can recognize it wherever we see something using it. Until now, artificial intelligence solutions seem to be machines, applications and algorithms trying to replicate or simulate human intelligence.

But is it “intelligence” that moves us to repeat a task that we have done before? Is repeating familiar tasks a sign of intelligence? Or intelligence is just involved when we do something for the first time? And then comes the repetition until automation, things that can be perfectly done by anybody or anything else… even a machine!

I really don’t feel that my intelligence is being part of the equation while doing my daily routine or tasks such as walking or talking, but now there are a lot of people that think machines have crossed the line because they can generate text or images.

It seems like everything was under control when we saw machines replicating movements, doing calculations or evaluating all possible combinations to find the best way of playing a game... but now, they can generate new things and it doesn’t mean that we have to fear them! Quite the opposite, we can use them in more fields to empower ourselves, making more things in less time.

“Generate” is not synonymous of “Create”

Even though the words may sound similar, there is a difference between them that is very important to understand when we talk about AI. Generative AI, as we can imagine, can generate things such as text or images based on other existing texts or images. We can use an AI tool to generate a picture of a turtle with butterfly wings, that is something that doesn’t exist in reality and probably our AI tool hasn't “seen” before. But the existence of turtles and butterflies enable the generation of the new creature.

“ It seems like everything was under control when we saw machines replicating movements, doing calculations or evaluating all possible combinations to find the best way of playing a game... but now, they can generate new things and it doesn’t mean that we have to fear them”

To create something means to define something undefined, and that is impossible for AI algorithms. When a user is writing a prompt, the way of talking to this kind of algorithms, explaining what needs to be written or drawn, the user is going through the process of creation. The AI is just a tool, like a hammer or a screwdriver, that enables human beings to materialize their creations faster.

How do we know it is time to use AI?

Every time you feel you are not adding value to a task, it is time to delegate it to a machine. Of course, if that algorithm already exists.

For example, AI tools can read documents and create summaries or translate them into almost any language, answer questions using the information contained in different documents. There are also models that can analyze images and describe them or detect patterns, check if something appears or not in that image. And perform any of those tasks in milliseconds. It’s not worth spending time doing things while they can do it for us.

They learn from our data

All these solutions need to be trained, and this is the point where humans can infect machines with their biases. If we don’t give machines enough diversity on the data that is used to train the models, and all possible cases are not being “shown” to the machines, it is obvious that the outputs we can get from them will not cover all the cases or situations; remember that they can’t create new things, just generate combinations of existing things.

So, before saying that an AI algorithm is making some kind of mistake, make sure that it has been trained properly. Perhaps it just needs more training and learning from more and more data.

Conclusion

“In summary, Artificial Intelligence has a long history and is now an integral part of our daily lives. While questions persist about its replication of human intelligence, AI's potential to empower us is evident. Recognizing the difference between "generating" and "creating" is crucial, and we should delegate tasks to AI when we add little value. However, vigilance against biases in AI training is essential for accurate outcomes. As AI continues to evolve, responsible and thoughtful use will define our collaborative future with these technologies.”

This last paragraph was written by an artificial intelligence algorithm, based on the previous text. It’s kind of fun reading the words “us” and “we” (which, in this context, refer to humans) on a machine generated text. But I’m ok with it!