GPT-3, the new version of the OpenAI API, brings us closer to what until recently we could consider as science fiction. And it is that since the Californian company announced, in June of this year, the release of its beta version, it has done nothing more than generate a stir and expectation.
Many Twitter users did not know if the articles or tweets they were reading had been the work of humans or a computer program.
In 1953, the English writer Roald Dahl published a short story called The Great Automatic Grammarizer.
In it, Adolph, a frustrated brilliant engineer for a major company, creates a machine that allows you to write stories with the desired plot and style. According to his account, 5,000 words in just 30 seconds. With this invention, he and his boss, Mr. Bolhen, intend to corner the market in the best literary magazines and earn a lot of money.
Until very recently this was something reserved for science fiction.
But…
What would Dahl think if he saw the progress he and natural language processing (NLP) have you had in recent years?
What is GPT-3
GPT-3 is a natural language model that allows you to generate automatic texts of any kind, from poetry to legal, economics or politics. A cloud-based AI platform that enables developers to build and train machine learning models. GPT-3 offers a comprehensive set of APIs for natural language processing, allowing developers to build applications that understand human language.
On many occasions, it is not easy to distinguish the result from that of a person of flesh and blood. In fact, the last two sentences of this first description have been created by GPT-3 itself, hadn’t you noticed?
But, don’t just be afraid of the editors! This new technology also performs calculations, writes guitar tablature or answers complex questions, without being trained in advance. I could even prepare for write code in various types of computer language such as JavaScript or HTML, translate from one to another or from natural to computer language.
Get ready Here’s something for everyone!
GPT-3 has read everything from news, blog articles, to online encyclopedias
Already the release of GPT-2, its previous version, caused a stir by being classified by its creators as dangerous to be used for unethical purposes.
Its potential to generate large amounts of text in a matter of seconds made it possible to use it to create , increase disinformation, etc. Although, after a few months, and saying that perhaps they had exaggerated a bit, they decided to put it on the market.
But it is that GPT3 is not just a little more powerful; while GPT-1 and GPT-2 worked with 110 million and 1.5 billion parameters respectively, GPT-3 works with no more and no less than 175 billion.
This new model has absorbed more than 400,000 million texts present on the network. It is then through these texts and knowledge that GPT-3 is able to generate coherent stories or essays and predict the meaning of the text based on certain data.
But, unlike what we might imagine, this technology does not read information like we do, from left to right of the sentence, nor, like in other countries, from right to left, bottom to top or top to bottom. But we could say that it reads everything at a stroke and instantly, in just seconds.
Oh, if only we could have read all the school lessons like this!
Not that this technology in shop what he reads or writes in the human sense of the term, but, so to speak, is capable of generating coherence and getting it right based on a series of mathematical parameters based on the enormous amount of information absorbed.
Microsoft bets on GPT-3
The possibilities posed by this model have not gone unnoticed by Microsoft, which already mid 2021 announced at its Build developer conference the first integration of GPT-3 in Power Appsyour application development platform.
Thanks to this union of forces, Microsoft intended bring the creation of apps closer to everyone, eliminating the need to understand the language in code and know the programming formulas. Both expert developers and people with little or no experience could develop applications to boost the productivity of their businesses.
Azure OpenAi Service, the union between the language models of GPT-3 and the business capabilities of Azure
months later, in November 2021, Microsoft introduced new advances and improvements for their implementation of GPT-3 in their products during their Ignite conference. These developments focused on the new Azure OpenAI Service, which allows access to the OpenAI API through the Azure platform. In this way, Microsoft customers gain access to OpenAI’s GPT-3 models, combined with the enterprise-grade security, privacy, compliance, and other capabilities offered by Microsoft Azure.
Eric Boyd, Microsoft corporate vice president for Azure AI, said that this is just the beginning and that, for now, they have only slightly scratched the surface of the possibilities that GPT-3 can offer. “We are only in the beginning stages of discovering what the power and potential of GPT-3 is, which is what makes it so interesting. Now we’re taking what OpenAI has released and making it available to all the business promises that companies need to go into production.”.
Likewise, new tools were also announced available to clients, which will serve to ensure that the results of the model are suitable for their businesses. In addition, how this technology is used will be supervised, so as to ensure that it is used correctly.
Dominic Diva Karuniproduct manager of the Microsoft division that leads Azure OpenAI, revealed some potential uses of this new product: “It helps speed up the creative writing process, helps them extract information from a large amount of text, and its code generation capabilities are a great example of the new kinds of business value these models bring. Clients learn new things about what can enlighten them every day.”
From the creation of websites to the optimization of financial services, other tools that already use GPT-3
As in the case of Microsoft, other companies and solutions have decided to boost their activity with the wonders that GPT-3 puts at their disposal. These are some examples:
This tool makes it easy to create applications and web pages, speeding up the process while maintaining efficient results thanks to GPT-3 technology. It also allows you to create SQL codes, React components and visually assemble an interface.
This omnichannel communications platform has integrated GPT-3 technology to optimize customer interactions, authentications and video calls for marketing and sales companies.
This tool provides administrative and financial services to bars and restaurants. Thanks to GPT-3 you can combine the latest technology with the work of professionals, offering improved accounting, payroll, analysis, taxes, etc. services.
This solution uses GPT-3 to automate text creativity tools. With Copy.ai you can generate copy for marketing campaigns, networks, websites, blogs, etc., with a single click.
But is it really that amazing?
As its own creators affirm, . It still makes mistakes, and sometimes has inconsistencies.
In addition, being based on texts and material created by human beings throughout the world makes it assimilate many of the biases and prejudices. There have been examples where even GPT-3 responds in a racist or sexist way to certain questions.
However, the fact that it is capable of performing specific actions in natural language processing, without having been specifically trained to do so, represents an important step in .
In view of the rapid advances that are taking place in this field, in the coming years we will undoubtedly be forced to rethink our way of understanding and interacting with the world around us, especially with technology.
In Roald Dahl’s story, Adolph’s story-typewriter looked like the dashboard of an airplane. Pressing a series of buttons or levers preselected the plot or style, category, theme, characters, or lexicon. Who can assure us in a few years that an article like this has not been written with buttons and levers by the prolific author named GPT-X?
For now, We leave you this wonderful conversation between and Marc about it generated by GPT-3:
Elon Musk: Hello Mark. How are you?
mark zuckerberg: Fine, Elon. Thanks for asking. And you?
Elon Musk: I’m fine too. I’m glad we were able to see each other in person again. It’s been a while since we last saw each other in the Metaverse.
mark zuckerberg: Yes, it’s been a bit of time. But I think this new virtual world is worth exploring together.
Elon Musk: I’ve been thinking about the metaverse lately and wanted to share my thoughts with you.
mark zuckerberg: Sure, I’d love to hear what you have to say.
Elon Musk: Well, as you know, the metaverse is a simulation of reality built by computers. And I think it could be very useful, but also dangerous.
mark zuckerberg: Dangerous? In what sense?
Elon Musk: Well, if people spend too much time in the metaverse, they could lose touch with reality and live in a virtual world. And that could be very dangerous.
mark zuckerberg: Hmm, yes, that’s true. But I think it could also be very useful for people. For example, we could use the metaverse to help children access education around the world.
Elon Musk: Yes, that’s true. But I also think the metaverse could be used to control people.
mark zuckerberg: Hmm, yes. But that is not its purpose, with the Metaverse we want to help people.
Elon Musk: Yes, but I think we should be careful with the metaverse. I don’t want people to lose touch with reality and live in a virtual world.
Stay informed of the most relevant news on our news channel
Facebook Comments
Disqus