What exactly is ChatGPT? I asked the buzzy artificial intelligence chatbot, which has sparked debate in classrooms, corporate boardrooms, and on social media, to explain itself.
ChatGPT is “an AI-powered chatbot built by OpenAI, based on the GPT (Generative Pretrained Transformer) language model,” according to its own description. It use deep learning algorithms to generate conversational answers to text inputs.”
The tool is making waves in the business sector. Management from a variety of companies, including oil giants, banks, and even industrial powerhouse Caterpillar, have referenced it on earnings calls.
It has also raised concerns about possible abuses. ChatGPT has been used in classrooms to produce whole essays, and hackers have begun testing it to develop harmful code.
So, what exactly is ChatGPT? Here’s a quick rundown of everything you need to know about the popular AI chatbot.
What exactly is ChatGPT?
OpenAI’s ChatGPT is an AI chatbot developed in San Francisco. Elon Musk and Sam Altman co-founded OpenAI in 2015, and it is supported by well-known investors, most notably Microsoft.
It is one of several generative AI examples. These are technologies that allow users to provide written prompts and receive fresh human-like text, photos, and videos generated by artificial intelligence.
Previous examples include Dall-E, an OpenAI text-to-image software that attracted people with its capacity to generate realistic, often ludicrous, images that fit people’s written descriptions.
Lensa, an app built on the open-source AI project Stable Diffusion, has been used to transform selfies into illustrious self-portraits inspired by sci-fi and anime.
In the case of ChatGPT, a service is a text-based tool that can generate human-like responses to user requests ranging from William Shakespeare-style poetry to suggestions on what to do for a child’s birthday party.
What makes it so unique?
ChatGPT is powered by a large language model, also known as an LLM, which is built to interpret human language and generate responses based on vast data corpora.
GPT-3.5 is the name of ChatGPT’s LLM. It is an enhancement to OpenAI’s GPT-3 language model.
GPT-3 is one of the largest and most powerful language-processing AI models to date, with 175 billion parameters.
What makes ChatGPT so astounding is its capacity to generate human-like responses, thanks in large part to the massive volumes of data on which it is taught.
“What’s interesting is that the replies are becoming more and more human-like, so you’re seeing things that we didn’t think computers could do previously,” said Jeffrey Wong, global chief innovation officer at professional services firm EY.
Another feature that sets ChatGPT apart is its ability to log context from previous messages in a thread and use it to create responses later in the conversation.
What makes it so popular?
No other generative AI application has achieved the level of influence and virality that ChatGPT has.
It was the subject of numerous memes and the talk of the corporate community during last month’s World Economic Forum in Davos, Switzerland. Baidu, the Chinese tech titan, created its own version named Ernie Bot.
According to Altman’s Dec. 5 tweet, the chatbot signed up 1 million people in the five days following its launch. According to a UBS note published last week, ChatGPT had 100 million monthly active users barely two months after its inception, making it the fastest-growing consumer application in history.
TikTok took nine months and Instagram two and a half years to achieve 100 million members.
According to Similarweb data, January 31 was the busiest day ever for ChatGPT, with its website receiving a record 28 million daily views. This was an increase of 165% from a month ago.
One reason for ChatGPT’s popularity is its ease of use. Anyone can access the service through the OpenAI website, and its possible applications vary from school assignments to legal papers.
According to Wong, timing has also played a role.
“You often see this explosion of the invention when we come out of pandemics,” he remarked. “The most notable example is that following the Black Plague, there was this Renaissance, this surge of innovation all throughout.”
Why are tech titans at odds?
Microsoft is putting billions of dollars into ChatGPT’s owner, OpenAI. The internet titan signed a multiyear, multibillion-dollar investment deal with OpenAI in late January.
Microsoft declined to provide a monetary figure. According to Semafor, the Redmond, Washington-based computer behemoth is in talks to invest up to $10 billion in the startup. Microsoft previously made a $1 billion investment in OpenAI.
Microsoft held a press event on Tuesday to unveil new AI-powered improvements to its Bing search engine and Edge browser. Altman acknowledged that parts of OpenAI’s GPT-3.5 language technologies had been integrated into Bing.
That was a day after Google announced Bard AI, its own response to ChatGPT. Bard will be available on Google Search in the next weeks, according to the business.
Google is concerned about ChatGPT. People could rely on ChatGPT instead of the web search pioneer for their most pressing questions.
Google was actually a forerunner in the advanced conversational AI game, launching its own massive language model dubbed Lamda in 2021. It missed the boat by not introducing its own consumer product based on Lamda — something it hopes to rectify with Bard, which is powered by the language paradigm.
How effective is ChatGPT?
ChatGPT has restrictions. The chatbot’s responses may contain factual errors. It can, for example, create false historical names and publications that do not exist, or it can fail to solve certain math problems.
“These models’ open-ended elements are a double-edged sword,” Will Williams, vice president of machine learning at British AI firm Speechmatics, told CNBC.
“On the one hand, there is a great amount of flexibility and fluidity in interactions, and entertaining discourse on practically any topic is feasible. On the other hand, you never know when the model is in touch with reality and not just hallucinating.”
ChatGPT’s expertise is still limited to data from 2021, although it may improve over time. ChatGPT is expected to be the prelude to considerably more complex AI systems in the future.
For the time being, researchers believe generative AI is incapable of obtaining human-like, “general” intelligence.
Artificial general intelligence, or AGI, is frequently regarded as the AI community’s holy grail. It most typically refers to an intelligent agent’s ability to comprehend or learn any intellectual job that a human being can.
Many businesses, from OpenAI to Google’s DeepMind, seek to achieve this.
GPT-3’s capabilities have already sparked interest in OpenAI’s next-generation LLM model, GPT-4.
To temper expectations, OpenAI’s Altman recently told StrictlyVC that people were “begging to be disappointed” by the buzz around GPT-4.