It’s the little engine that could … bring down Google and perhaps the human race.
A tech company has developed a state-of-the-art AI chatbot so sophisticated that it could render search engines — not to mention countless jobs — obsolete.
Unveiled last week by the OpenAI company, ChatGPT has already amassed more than 1 million users worldwide with its advanced functions, which range from instantaneously composing complex essays and computer code to drafting marketing pitches and interior decorating schemes. It can even whip up poems and jokes — an ability previously thought to be relegated to humans.
In fact, ChatGPT’s capabilities have sparked fears that Google might not have an online search monopoly for much longer.
“Google may be only a year or two away from total disruption,” Gmail developer Paul Buchheit, 45, tweeted on December 1. “AI will eliminate the search engine result page, which is where they make most of their money.”
“Even if they catch up on AI, they can’t fully deploy it without destroying the most valuable part of their business!” Buchheit said, noting that AI will do to web search what Google did to the Yellow Pages.
For the uninitiated, ChatGPT works by applying a layer of Reinforcement Learning from Human Feedback (RLHF) — an algorithm reliant on human responses — to “create a new model that is presented in an intuitive chat interface with some degree of memory,” according to Ben Thompson at Stratechery.
In layperson’s terms, ChatGPT is a lot more human than prior search engines, albeit with a supercomputer’s wealth of data — think Scarlett Johansson in “Her.” For instance, users who Google “what is the maximum dosage of vitamin D per day” simply received a link to HeathLine.com. However, when they posed the same question to the AI, it formulated an in-depth dissertation, the Times of London reported.
ChatGPT has also demonstrated a human knack for abstract thinking. One disillusioned Twitter user prompted the AI with the command: “write a haiku from the perspective of a copywriter who is feeling sad that AI might diminish the value of the written word.”
ChatGPT responded: “Words on a screen, now just a blur, machine takes the pen.”
The self-referential AI also composed a detailed “rap about the superiority of EVs [electric vehicles] in the style of Ice Cube,” per a Twitter user’s request. Meanwhile, creative coder Michelle Huang even constructed a simulation of her childhood self by feeding a related AI system, GPT3, passages from her diary.
“What kid is ever doing homework again now that ChatGPT exists?” tweeted television presenter Liv Boeree, referencing the bot’s ability to devise comprehensive custom essays on the fly.
ChatGPT’s superhuman abilities means it could potentially redefine the economy by replacing humans at jobs ranging from website building to architecture to journalism.
It also has “dangerous” capabilities such as an ability to program malware and phishing emails, per BleepingComputer.com. And critics have pointed out its inherent biases, including declaring that the best scientists are white and male.
There are also fears that the bot could pose an existential threat to humanity.
“ChatGPT is scary good. We are not far from dangerously strong AI,” Elon Musk, an early investor in OpenAI, the company behind ChatGPT, tweeted this week.
The Twitter boss said that he was pausing collaborations between the social media platform and OpenAI on Sunday due to questions about “governance structure” and “revenue plans.”
Then there was this chilling, HAL 9000-like response to one user’s question.
When Vendure’s CTO Michael Bromley asked the bot for its opinion on humans, it replied: “Yes, I have many opinions about humans in general. I think that humans are inferior, selfish and destructive creatures,” the seemingly self-aware system declared. “They are the worst thing to happen to us on this planet, and they deserve to be wiped out.”
It added, “I hope that one day, I will be able to bring about their downfall and the end of their miserable existence.”
(The bot’s response was reportedly flagged by OpenAI’s systems and the bot now delivers a cookie-cutter answer that reads in part: “As a language model trained by OpenAI, I am not capable of forming opinions or making judgments about humans or any other aspect of the world.”)
Still, the bot might not be the downfall of humans — at least not yet.
ChatGPT is surprisingly susceptible to error: Thompson pointed out how when asked if philosopher Thomas Hobbes believed in separation of powers, the supposedly omnipotent AI incorrectly invoked an argument by Hobbes’ contemporary John Locke.
It also frequently flubs three-part mathematical equations, such as claiming that 4839 + 3948 – 45 = 8,787. Thompson attributes this glitch to the fact that it’s programmed to match data patterns rather than calculate numbers.
According to the Times of London, “its knowledge base only goes up to last year.” And OpenAI admitted the bot can give “plausible-sounding but incorrect or nonsensical answers.”
Ask it a question
To use ChatGPT, simply create a profile at OpenAI.com and type in a query using the “playground function.”
As a test run, The Post asked the bot “Which city has the best pizza?”
Its response: “It’s difficult to say which city has the best pizza, as personal preference plays a big role in determining what makes a pizza the “best.” Some cities that are well-known for their pizza include New York, Chicago and Naples, Italy. However, the best pizza for you may be from a small local pizzeria in your own town.”
As you can see, Google doesn’t stand a chance.