The Future is here (or at least a good preview of it).
We couldn’t start talking about ChatGPT any other way.
This name, heard more and more wherever you go, especially in the science and technology area, will become more and more common - in all areas. And you will end up using it - like it or not. But don’t worry, you will probably like it.
And even though you may not like it, it will be part of your life.
Just like Google, as an Internet search engine, is part of your life today.
Yes, ChatGPT has huge potential to trigger changes as big - or even bigger - than what we’ve had in the way we search the Web (before and after Google).
That scene from the Terminator movie, where they try to turn off the machines, is no longer “something almost impossible”. Now, it is at least a “remote possibility” in the future.
Until recently, everyone understood artificial intelligence as a bunch of IFs and ELSEs. And ready. The same way it was “programmed via code”. Which is still true, as a basic concept.
What happens is that the logics evolved - and continue to evolve. Algorithms are increasingly complex, taking advantage of the huge amount of available data and computational infrastructure to process it.
5G is here, not with the aim of a “transformation” in modern communication, but instead a “revolution”. Its successor, the 6G, will follow the same path.
And ChatGPT is part of this revolution!
So let’s understand a little bit about it.
ChatGPT was launched on November 30, 2022 by American company OpenAI.
It is a prototype of an artificial intelligence chatbot. A robot.
Until recently, chatbots were known for having a limited dictionary and the memory of a fish (like Dory from the Disney Pixar animation, Finding Nemo, who had memory problems) that forgets everything in a few seconds.
This is actually a myth, because studies have already shown that fish do have memory, which can last for years, and also that they have excellent learning ability!
As well as the demystification of fish memory, as well as the giant and exponential expansion of limited dictionaries, came ChatGPT, reacting more coherently according to the history of searches and responses from a vast source of available information processed by complex algorithms.
The company that created ChatGPT itself says: “The dialog format allows ChatGPT to answer specific questions, admit its mistakes, question incorrect assumptions and reject inappropriate requests”, which actually translates into a more user-friendly artificial intelligence system, being able to talk, write texts and even produce videos and images from a vast database of digital books, online publications and other media - as if you were really talking to another person, a written dialogue with the person who is asking questions.
It works! It really gives the impression that it actually “think”, showing empathy, building relationships and speaking correctly.
“Going viral” a few days after its launch, ChatGPT has already been tested by millions of users, showing that artificial intelligence has become popular.
It’s important to point out here that ChatGPT is not the only one. As expected, there are (many) very good “competitors” or similar ones, such as Cicero from Meta company (Facebook) or Character AI (from a company founded by former Google engineers).
Technologies based on Artificial Intelligence are increasingly popular, especially in recent months, with significant launches using this technology.
ChatGPT stands out among them perhaps because it integrates a wide range of technologies developed by OpenAI and because it has a close relationship with Microsoft. (for example, Windows Operating System has a market share of around 90% in the World).
But of course, there are issues. Or “challenges” to be overcome.
While ChatGPT is notable for its detailed and articulate responses, the accuracy and use of its information is the subject of much debate.
ChatGPT itself informs in its limitations that: its generated answers may be incorrect, that harmful instructions or biased content may be produced and also that it has limited knowledge of the world and events after 2021.
As for limited knowledge, this will probably cease to be a problem soon: just connect to the Internet and update your base almost instantly.
What worries me most are the generated responses, as they are “predictions” based on a “trained model”.
Let’s take a short break, and understand what this means by “trained data models”.
Most “big data” data, such as that present in large companies and on the Internet, is “unstructured” data, and therefore traditional algorithms are not able to analyze it. The big challenge is to figure out how to process and use all this data.
And that’s where Artificial Intelligence comes in, more specifically Machine Learning.
We won’t go into detail about this today, but with “historical already classified data” (by history, considering already known conditions and values) we feed a “machine learning algorithm” (which will learn which conditions are necessary for each classification). The key to artificial intelligence is well-trained models.
That way, when we have “new data”, the “trained model” will be able to make predictions.
Now let’s get back to the ChatGTP case.
ChatGTP models were trained on large amounts of “human-written” Internet data, including conversations, so the responses provided can appear human.
It is important to understand, therefore, that this is a direct result of system design (ie, maximizing the similarity between the outputs and the dataset on which the models were trained). And that the result (the outputs) can be inaccurate, false and misleading at times.
In other words…if the sentences and citations (data) are correct, the result (output) will be correct.
Let’s use an example to make it simple. Imagine that you don’t know the color of the grass.
And decide to do a search on the Internet.
You’ll find thousands and thousands of websites, almost all written by humans.
You will find this answer on biology, science and even song lyrics websites!
After reading some of these definitions (of the color of the grass), your “brain algorithm” in your head you already have the definition: the color of the grass is green!
But of course, there can always be mistakes: people who wrote that the grass is red, or yellow. However, these cases are not the majority. Doing a simple algorithm we can get the highest amount of response, and depending on this percentage, we can say that the grass is green.
Although ChatGTP’s algorithm is not as simple as our example, it follows the same logic.
ChatGTP was “born” with OpenAI’s GPT-3.5 language model, and was “educated” using supervised learning and reinforcement learning (training machine learning models to make a sequence of decisions).
Then using an autoregressive language model, it achieves deep learning to produce human-like text.
Basically you ask and it answers. Given an initial text as a question (prompt) it will produce a text according to what was asked (following that prompt).
Prompts are the foundation of any Artificial Intelligence tool. And of course, that includes ChatGPR.
To understand the importance of prompts, let’s use one sentence. Some say it was attributed to Albert Einstein, but it seems that he didn’t said it; Anyway, here’s the sentence:
“The definition of insanity is doing the same thing over and over and expecting different results.”
This perfectly applies to ChatGPT (and Artificial Intelligence systems).
In Artificial Intelligence systems, the “better” (detailed) your request is, the better you will be served.
In ChatGPT’s case, if your question is good, your answer will be even better.
Take the test: for example, ask what is the salary for your profession. ChatGPT will answer.
Now, ask ask the same question, but with extra detail. For example: ask the salary of your profession in your State, for a professional with 20 years of career. ChatGPT will answer better (a more specific answer).
Another good tip is: if you are not satisfied with the result, ask him for additional suggestions/information.
Another good tip is: if you are not satisfied with the result, ask it for suggestions/further information.
Don’t be afraid, use your creativity, and explore a lot, don’t get caught up in trivia.
ChatGPT has gone viral, it’s the hype. And a lot of people end up writing superficial guides and tutorials, often without any really useful information. Most seek only to teach “tricks”, not really teaching how to take advantage of the full potential of one of the most powerful tools of all times.
Then change it. Really really use ChatGPT in the most effective way possible. We’re talking here about a superpowered computer that follows a giant list of instructions on what to do in each situation. So it’s up to you to extract as much as you can.
And now let’s get back talking about some more challenges ChatGPT has.
In addition to the problem of data accuracy, another problem that has been plaguing quite a few professionals is the issue of unemployment that ChatGPT (and of course other AI applications) can generate, since so many of these tools can replace human activities. This uncertainty of the future can arouse a certain discomfort.
Many cite, for example, the example of workers in a farm now being replaced by more “efficient” tractors and agricultural machinery. Nowadays, the professions most concerned about losing their jobs in a few years are especially professionals such as teachers, programmers and journalists; but the concern is general and in all areas - including Telecom and ICT professionals.
What we need to remember is that all the time and everywhere new AI applications appears, in the most different areas (science, technology, security, marketing, etc.).
But we must not be afraid, nor frightened.
Quite the opposite, we must adapt to these innovations and technologies, understand how they work and take advantage of it!
Even ChatGPT admits to some of the risks involved and gives some advice to users - like this response it gives when responding to a question about concerns related to the unemployment trend:
“Develop new skills and competencies. As AI technology continues to advance, many jobs and industries are likely to be affected. To be more competitive and relevant as a professional, it is important to develop new skills and competencies that are in demand in the AI-driven economy.
This can include technical skills like programming, data analysis and machine learning, as well as soft skills like critical thinking, communication and collaboration.”
If you don’t understand, it means something like this:
Artificial Intelligence is not going to steal your job. But a person who is interested and who knows how to use it (Artificial Intelligence) in their favor, yes!
Don’t waste time, start entering this world. Seek help - and always remember that we at the telecomHall community are always here, with a hand outstretched to help you!
ChatGPT is here to stay. Artificial Intelligence is here to stay! And everyone needs to adapt to the new times.
Again: people won’t lose their jobs to AI, but they will lose them to people who use AI to their advantage!
Going back to the challenges to be overcome by ChatGPT, we still have the accusations of using its services for academic plagiarism and also lack of transparency.
For example, by not telling which sources ChatGPT uses.
When someone writes a book, bachelor’s or master’s thesis, etc. is obliged to indicate the sources used. But this does not happen in ChatGPT.
That is, ChatGPT is a black box, which you simply “have to believe”. And that doesn’t give confidence.
The generated texts are really “impressively detailed” and very “human-like”, and this can cause them to be used wrongly, for example in the academic world (schools and universities), where students will “stop thinking”, stop to produce… and they will “start copying”!
With no doubt, it is an issue that will be much discussed from now on.
Unfortunately ChatGPT cannot inform its “sources”, as the result is a probabilistic function of millions of citations.
It is only left to those affected, such as education providers, to adapt to improve critical thinking or reasoning. Even fighting with your own weapons: using a “ChatGTP detector!”
Finally, a last point that also attracts a lot of attention is cybersecurity.
ChatGPT is capable of a lot of things, much more than we’ve talked about so far.
For instance, it has been used by programmers to find bugs and fix code!
Yes: and impressively!
But this also shows that ChatGPT is capable of writing phishing and malware emails, for example when combined with another artificial intelligence model developed by OpenAI itself - Codex (which analyzes natural language and generates code in response).
With no doubt, this advance in Artificial Intelligence (not only with OpenAI applications, let’s be clear!) can pose a major cybersecurity risk.
But then again, all of this will be discussed further, that’s for sure. But beyond that, we’ve come to the end of our brief introduction to ChatGPT.
But we may still have a question: what does GPT mean (or what is the Full Form of GPT)?
How about we ask ChatGPT itself?
Here’s the answer:
GPT stands for “Generative Pre-trained Transformer.” It is a type of language model developed by OpenAI that uses deep learning to generate human-like text. It is pre-trained on a large corpus of text data and can be fine-tuned for specific tasks such as language translation, question answering, and text summarization.
It looks like the answer was written by a human. But actually it was a robot who wrote it.
CHatGPT is one of the most powerful tools of all time - and it’s up to you to take full advantage of the potential it offers, for you and your career!
Not only chatbots (such as ChatGPT), but also robots and androids are evolving a lot, becoming increasingly present in our lives, such as at the reception of a hotel, as a nurse for the elderly or as an employee in a Factory or House.
The only doubt then remains: will the day come when we will have to “turn off the machines”?
Jokes aside, enjoy the potential that ChatGPT has! Take advantage of the potential that Artificial Intelligence has! Leverage your career, your life.
And get started today by actively participating - helping to build our collaborative learning system.
Participate in the community, in the forum, by liking topics that have helped you. Ask when in doubt. Answer questions if you know.
Much greater than all Artificial Intelligence, the greatest asset we have, deep down, is us, our Community. Thanks for being a part!