What is the killer app for ChatGPT? Like all new technologies with widespread potential, this is the obvious question for an algorithm that can replicate human writing styles and give cogent, seemingly intelligent answers.
Right now, that killer app is asking ChatGPT to write a column or feature such as this to demonstrate how good it is.
This is reminiscent of the hunt for the killer feature for 3G, when the faster next-generation wireless broadband launched a decade ago and geeks and telecom engineers endlessly tested its speed with an app called Speedtest.net.
It might be condescending to call something as thoroughly good as ChatGPT a chatbot, but that is essentially what it is. Chatbot 2.0 is perhaps better.
ChatGPT is a hugely sophisticated chatbot. And, given that essays and other longer text are what it’s mostly being asked to produce at the moment, a sophisticated long-form chatbot.
That is not to say it isn’t groundbreaking and amazing. Soon after it was released on November 30 last year, it racked up 1-million users within five days. These were not geeks and technology journalists, but the average internet user discovering the joys of something remarkably good at mimicking we humans.
As renowned tech investor Paul Graham tweeted: “The striking thing about the reaction to ChatGPT is not just the number of people who are blown away by it, but who they are. These are not people who get excited by every shiny new thing. Clearly something big is happening.”
It is technically machine learning, where a software program is trained to train itself. As it processes data, the theory goes, it will get better at interpreting and learning.
With a new kind of chatbot technology poised to reinvent or even replace traditional search engines, Google could face the first serious threat to its main search business
Machine learning is effective in narrow-focused tasks, such as learning to play Go, the ancient Chinese strategy board game. Google’s AI Go project eventually beat the best human player, with a move that was described as “beautiful”.
ChatGPT is more sophisticated and shows what seems like an understanding of its subject matter. Made by a company called OpenAI, cofounded by Elon Musk, ChatGPT is the best example of what is known as generative AI.
One company already worried is Google, whose management declared a “code red,” the New York Times reported.
“For Google, this was akin to pulling the fire alarm,” the paper wrote. “Some fear the company may be approaching a moment the biggest Silicon Valley outfits dread — the arrival of an enormous technological change that could upend the business.”
Why scroll through a page of search results when ChatGPT might give a more specific result?
As The Times says: “For more than 20 years, the Google search engine has served as the world’s primary gateway to the internet. But with a new kind of chatbot technology poised to reinvent or even replace traditional search engines, Google could face the first serious threat to its main search business.”
The paper reported that one executive described the search firm’s response as “make or break for Google’s future”.
“Google has a business model issue,” former Yahoo and Googler Amr Awadallah told the paper. “If Google gives you the perfect answer to each query, you won’t click on any ads,” said Awadallah, who is building similar technology through his start-up Vectara.
Google has its own generative AI project called LaMDA, a language model for dialogue applications, which is apparently so good that last year one of Google’s engineers was fired after he claimed LaMDA was sentient.
Meanwhile, Microsoft has already announced it plans to augment its Bing search engine with ChatGPT. In 2019 the tech giant invested $1bn in OpenAI and it has reportedly invested $2bn more.
Last week Microsoft offered to invest a further $10bn for a 51% stake, according to online publication Semafor, which would have valued the San Francisco AI start-up at $29bn.
Last April another OpenAI project preceded this hype when it launched DALL-E, which could produce sophisticated illustrations from just a few typed prompts such as “a teddy bear playing a trumpet underwater” or “cats playing chess”. The results are impressive. One illustration even won a minor art competition, causing widespread debate about what constitutes art.
Arguably the most contentious application for ChatGPT is essay writing, where US universities have already picked up on plagiarism from using the chatbot. The simple giveaway, which should be obvious to those worried about this unethical use, is that a student suddenly submits an astoundingly good essay. Common sense dictates that the student might have had some help. Existing software products for universities to identify plagiarism are going to require an upgrade to detect ChatGPT’s superior output.
As Graham wrote, “clearly something big is happening”.
Would you like to comment on this article?
Sign up (it's quick and free) or sign in now.
Please read our Comment Policy before commenting.