What is GPT-3
GPT-3 or Generative Pre-trained Transformer 3 is a language prediction model created by OpenAI, an artificial intelligence research laboratory in San Francisco. It is the third version of the software. In layman terms, it is an algorithm that is a part of the deep learning section of Machine Learning that can generate text or language – summary, poem, essays, fiction – anything a human brain is capable of writing.
The algorithm has been pre-trained with almost 570GB of text data collected from CommonCrawl, Wikipedia, and OpenAI. With 175 billion parameters (learning rate, batch size, etc.,) it is nothing if not the most powerful language model ever built. When it came out in May 2020, everybody from tech researchers to college students wanted to get their hands at it. GPT-3 currently is in private beta and people can sign up on a waitlist to use it. OpenAI has offered it as an API accessible through the cloud.
Now, since GPT-3 is a pre-trained model, it simply requires few sets of inputs to produce an output. For example, Mario Klingemann, a German artist and Google Arts and Culture resident, fed the GPT-3 model with the title ‘The importance of being on twitter,’ the author’s name, ‘Jerome K. Jerome,’ and the first word ‘It.’ The model quickly produced a 2300+ word essay.
Once a request is entered (in this case, ‘The importance of being on twitter,’ ‘Jerome K. Jerome,’ and ‘It,’) the application analyses it and provides the most probable answer. The text predictor processes all of the text existing on the Internet, calculating the most statistically expected output. It can do this because of the 570GB data provided to it by OpenAI for it to understand how languages and sentences work and are written or structured. This training time is said to have cost OpenAI almost $4.1 million.
What can GPT-3 do?
GPT-3 made a lot of buzz as its API maintained by OpenAI was released. Let’s take a look at the most interesting use-cases of GPT-3:
1. Write Code:
In a first, Sharif Shameem, the founder of Debuild, used GPT-3 to write JSX code. He entered the input in plain English and GPT-3 produced JSX layouts in a matter of seconds!
On a different try, GPT-3 wrote SQL codes:
2. Create an app:
A YouTube video titled ‘GPT-3: Build me a Photo App,’ demonstrates feeding simple data into GPT-3, the output of which is Designer, an app that is an integration of Figma plugin and Instagram UI. Amazing!
3. Integrate into an app:
Microsoft recently launched its first features in a customer product powered by GPT-3 – in Microsoft Power Apps. Power Apps is a low code development platform to help everyone – from people with no coding experience to professional developers – to build applications. This integration will allow people to describe coding goals in simple English language to the computer. GPT-3 then transforms the command into a Microsoft Power Fx formula.
“It’s one of the first implementations showing how GPT-3, running on Microsoft Azure and powered by Azure Machine Learning and one of the first internal uses of its new managed endpoints capability, can solve real-world business needs on an enterprise-scale.” – Microsoft.
The most widely known and tried use-case of GPT-3. Look how The Guardian got itself a 2000 words article titled, ‘A robot wrote this entire article. Are you scared yet, human?’ written by GPT-3!
Who thought we’d have AI writing memes for us? Somebody on Twitter fed summarised version of some memes to GPT-3 and what it created can definitely be labelled as ‘funny:’
6. Business Ideas:
When GPT-3 was fed with a business list, the output was a bunch of new project ideas!
Which companies are using GPT-3 and how?
According to Open AI, companies have already started using GPT-3 to enhance their customer experience. For example,
1. Viable, a company that provides structured customer feedback to its own customers, uses GPT-3 to summarise feedbacks and make it accessible through simple English sentences. For example, if asked, “Why are customers irritated by the chatbot?”, Viable, using GPT-3, would provide the information: “Customers are irritated because the chatbot is taking too much time to reply.”
“GPT-3’s ability to identify themes from natural language and generate summaries allows Viable to give product, customer experience, and marketing teams at companies across industries a better understanding of their customers’ wants and needs.” – Daniel Erickson, CEO of Viable.
2. Fable Studio, a creative agency, is using GPT-3 to enable their ‘Virtual Beings’ (animated characters) to have natural conversations.
“GPT-3 has given us the ability to give our characters life. We’re excited to combine an artist’s vision, AI, and emotional intelligence to create powerful narratives and believe that one day, everyone will know a Virtual Being.” – Edward Saatchi, CEO of Fable Studio.
3. Algolia, an AI-powered search and discovery platform, has integrated GPT-3 into their advanced search technology to offer lightning-fast semantic search for their users. According to OpenAI, Algolia saw 91% more precision than BERT and could easily answer more complex queries.
“GPT-3 allows Algolia to answer more complex queries than ever before with our Algolia Answers product, identifying deeper contextual information to improve the quality of results and deliver them in seconds.” – Dustin Coates, Product and GTM Manager at Algolia.
4. Copy.ai provides AI-powered copywriting tools with the help of GPT-3. The most famous among its customers are social media post copy, blog content, website copy, sales and digital ad copy and product descriptions for e-commerce.
“Its ability to generate natural language that sounds like a human is uncanny and it’s by far the most powerful AI ever publicly released. We decided to build a project around copywriting using the AI service to create common marketing use case tools.” – Paul Yacoubian, Founder, Copy.ai
5. Latitude, a startup building games with infinite storylines is using GPT-3 to enhance its text-based adventure game, AI Dungeon. A player would type the action or dialogue they want their character to perform and using GPT-3 the game would produce the next phase of their personalised, never-ending adventure.
“How does the AI know what’s a good story? Because it has read a lot of good stories and knows the patterns involved in that.” – Co-founder and CEO Nick Walton.
6. Reply, an AI-powered sales engagement platform is using GPT-3 to manage hundreds of emails and contacts. It generates custom sales emails that look and feel as if they were written by a real person in seconds.
The list is endless. GPT-3 has revolutionised not just Artificial Intelligence but also the industries making use of it.
Anyone who has seen the results of GPT-3 knows that it is ahead of every language model we’ve ever seen. It is also no secret that the application has its limits and goes rogue when it comes to producing texts about topics revolving around racism, history, etc.
If you have any interesting ideas around GPT-3 or AI in general, let’s have a conversation. Feel free to reach out to us at +918002985878,+91439857338, or through the mail – email@example.com.
Alternatively, if you are thinking of leveraging the benefits of AI in your business, but confused about where to start, connect with us for a free consultation call – and we will help you with all the necessary details.