ADVERTISEMENT

Who’s afraid of ChatGPT?

Chatbots seem all set to steer the future as Artificial Intelligence of computers makes rapid strides to catch up with and perhaps overtake human intelligence. The Telegraph sits crystal-gazing at a New Town meet on AI

Sudeshna Banerjee Salt Lake Published 29.09.23, 11:15 AM
Sanjoy Guhathakurta, partner, IBM, addresses a session on AI and ChatGPT at Swapnobhor as organiser Pijush Ranjan Ghosh looks on.

Sanjoy Guhathakurta, partner, IBM, addresses a session on AI and ChatGPT at Swapnobhor as organiser Pijush Ranjan Ghosh looks on. Sudeshna Banerjee

Open your refrigerator, take a picture and upload on ChatGPT. It will list out what ingredients you have in stock, what cooking options you have with them, dig out relevant recipes from the Internet and even calculate the calorific value of each. All within seconds.

This wondrous natural language processing tool driven by Artificial Intelligence (AI) that can communicate in a human-like way through text messages can answer questions, generate articles, write essays, crack jokes, compose poetry, write code and even offer advice in response to prompts. No wonder, it has taken the world by storm.

ADVERTISEMENT

ChatGPT, the popular chatbot from OpenAI, a private company backed by Microsoft Corp, crossed the one million user mark in just five days after it was made public in November 2022. It went on to reach 100 million monthly active users in January, just two months after its launch, making it the fastest-growing consumer application in history.

In comparison, Facebook, one of the most popular social media apps, had taken around 10 months to cross one million users after its launch in February 2004. And Spotify, one of the most popular music streaming services, launched and reached a one million user base in March 2009, nearly five months after its launch in October 2008.

An interactive session was held at Swapnobhor senior citizens’ club recently for both elderly members with minimal idea about the world of technology and young and middle-aged professionals eager to know how the development could be put to use in their respective fields.

Boon or bane

The organisers set out to address the questions that the participants asked at the start of the session: Will it take away jobs? How can it help common people? How will it drive urban growth and rural infrastructure? What will be the skills of the future? How will it help specific fields, like advertising or music?

“We are trying to look into the future and predict what kind of disruptions can be expected and how it will impact the next generation,” said Pijush Ranjan Ghosh, a resident of Rail Vihar in Action Area 1 whose consulting firm Mindstrology had organised the programme.

The first question to ask, he pointed out, is whether ChatGPT or its counterpart floated by rival Google, called Bard, is a boon or a bane? Ghosh argued it was a boon. “The Industrial Revolution too came with machines that had taken away human jobs. What we need to see if, like machines, AI and its applications are increasing human efficiency. Some countries, like Italy and Spain, are trying to control its use. I think, like the internet, it cannot be controlled.”

No Generative AI tool, he clarified, can write poems like Rabindranath Tagore or make films like Satyajit Ray. “The concern is for the average workforce. Their quality will have to be improved to avoid being replaced by chatbots (‘bot’ being short form of robot, indicating a software programme that operates on the internet and does repetitive tasks). This revolution is like a challenge to all of us.”

ChatGPT, he added, needs to be taken seriously because so far all the text that chatbots used to generate read like they were indeed machine-generated. But ChatGPT has succeeded in passing the fine-tuning test.

Threats galore

The man called the godfather of AI, Geoffrey Hinton, who had worked on machine-learning algorithms at Google for over a decade, recently resigned so he could sound the alarm bells about the recent developments that he fears might trigger the “existential threat” of “super intelligence taking over from people”, to quote Hinton himself.

Once the use of ChatGPT becomes more prevalent, the first professionals to be hit would be content writers, script-writers and lawyers, Ghosh predicted.

Instructors at the University of Minnesota Law School had given the AI bot four examinations alongside real students and it earned C+, a low but passing grade. It performed better in essays than in multiple choice questions, especially those involving mathematics. Ghosh quoted a prediction that by 2029, 90 per cent scripts, movies and media content will be AI bot-generated. “Today a project takes 200-300 people, but by using chatGPT if the same job can be delivered by 10 people what will the others do?”

A similar dilemma was faced by Rolex, the Swiss watchmaker, in the 1980s when digital watches from Japan’s Casio flooded the market at far cheaper rates. Rolex rolled out a publicity campaign that stressed on it being a human labour-intensive product of precision and class. That helped in creating a different luxe segment for the Swiss watch. “Think of typists. Where did they go once computers came in? So generative AI will not replace us but will replace the work many of us do. The key word would be efficiency,” he said, emphasizing that there could be no going back on advancements achieved by this new technology.

OpenAI co-founder Elon Musk, Apple co-founder Steve Wozniak and other tech luminaries have signed an open letter seeking a pause on AI developments for six months, concerned at the race among AI labs to develop and deploy the technology without appropriate safety protocols.

Sanjoy Guhathakurta, a partner at IBM who holds four patents in AI innovations, pointed to education and health to be the two sectors to see the maximum disruption.

“An eye scan is enough for Google AI to predict if the person might suffer a heart attack. By 2040, there will be 33 million self-driving cars sold annually in the US. While driverless cars are unlikely on Indian streets in the near future, some areas here too will be affected,” said the resident of Greenwood Park, referring to chances of students turning to ChatGPT to produce essays.

“But I expect software to detect whether a piece of writing is human-generated or robot-generated will also be developed by then. Regulatory changes will also be needed if doctors, nurses and pharmacists have to be replaced by bots.”

The reason why lawyers are expected to be replaced to a large extent, Guhathakurta said, is because 95 per cent of their job is documentation and only five per cent is fighting cases in court. He conceded that initially there could be mistakes and the robotics engine could supply wrong data. “But it will become perfect over time. Our work pattern will change and manual labour will be reduced.”

Ghosh said he expects 90 per cent of movies in future to be created by AI. In daily life, ChatGPT will help us prepare CVs with correct formatting, draw up contracts and write emails. The problem of deepfake AI, a type of artificial intelligence used to create convincing image, audio and video hoaxes, will also surface.

While ChatGPT is good at mimicking conscious intelligence, the realm of the sub-conscious is still beyond it, the speakers pointed out.

Hidco managing director Debashis Sen said he had reached out to ChatGPT when residents sought an advisory in the aftermath of a child falling from the verandah last April. “It generated six pieces of advice. One was to make the grilles vertical and not horizontal so that they could not be climbed. I used that,” he said.

Sen referred to Israeli historian Yuval Noah Harari who has pointed to two capabilities that form the basis of human civilization — story-telling and speaking in different languages. “While a computer can switch from one language to another without having to be taught, now AI has hacked the operating system of the human civilization by coming up with story-telling computers, Harari has written.”

ChatGPT can be an excellent companion, Sen said. It can even take routine decisions like what to cook using a predictive algorithm.

Ghosh said an Apple watch on the wrist will replace the local doctor as it can read several parameters and issue warnings if it notices abnormalities. With drones coming into play, all deliveries will be at the doorstep. “More one-man companies will be registered as functions can be streamlined with generative AI and AI tools,” he said.

Ghosh demonstrated how ChatGPT works by logging in to https://chat.openai.com and asking it to write a paragraph on the event by typing in basic information. Within seconds, the programme generated a paragraph in correct and lucid English. He then typed in instructions to produce the same information in a point-by-point format and as an email. The chatbot promptly responded.

Guhathakurta ended by referring to a Coca-Cola ad launched this year. “It has characters coming out of iconic paintings in a museum and playing catch with a Coke bottle. It is a product of live-action images, digital effects and generative AI. If it has attained this level of perfection already, think of what is coming in the next two to three years!” he wondered.

Write to saltlake@abp.in

Follow us on:
ADVERTISEMENT