chatgpt disinformation: AI chatbots like Bard, ChatGPT stoke fears of disinformation nightmare

[ad_1]

Huge investments in pre-trained generative bots like Microsoft-backed ChatGPT and Google’s Bard are stoking fears of a disinformation spike, according to industry insiders. Social media intermediaries will also struggle to identify “fake” content and stop its spread in time, they added.

India, which is among the largest data markets in the world, is already battling a wave of disinformation compounded by the multiplicity of languages. Generative AI technologies may further exacerbate the problem according to technologists tracking the rise of these applications.

“The potential for disinformation is huge because these AI large-scale language models aren’t designed for factual accuracy, they’re designed for eloquent conversation,” Simon Greenman, co-founder, partner and CTO at, told ET. AI Best Practices.

Noting that AI’s large language models cannot distinguish between “evidence-supported information and fiction,” Greenman said “the fear we have, from a social perspective, is that it may amplify toxic content, racism, violence , misogyny, hate speech, and political theories that are flawed and biased.”

ChatGPT, created by SanFrancisco-based Open.AI, managed to reach 100 million users within two months of its launch.

Read also | OpenAI launches ChatGPT Plus: check price, features, details of the disruptive AI chatbot

Discover the stories you care about In January of this year, Microsoft announced a new multi-year, multi-billion dollar investment to underpin the technology. Microsoft declined to give details of the deal, but Semafor said Microsoft was in talks to invest up to $10 billion. The deal marked the third phase of the partnership between the two companies, following earlier investments by Microsoft in 2019 and 2021.

Earlier this month, Google announced it was launching Bard, a conversational bot powered by a “lightweight version” of Google’s Language Model for Dialogue Application, or LaMDA. Alphabet CEO Sundar Pichai also announced that LaMDA technology will be integrated into Google’s core search engine business in the future.

Disinformation is not new. But Greenman said the fear with big algorithm-driven mathematical language models is that they suck up content onto the Internet and then regurgitate it.

“So it’s sucking up misogynistic content from some dark corners of the internet. It’s giving it back and amplifying it,” she explained.

Read also | Creepy, malicious, dangerous: why Microsoft tamed the new ChatGPT-style Bing

According to the National Crime Records Bureau (NCRB), fake news incidents and the circulation of rumors increased nearly threefold in 2020 compared to 2019. A total of 1,527 fake news cases were recorded in 2020, up from 486 cases in 2019 and 280 cases in 2018.

However, after the pandemic year, there was a 42% drop in the number of cases recorded according to NCRB data for 2021 where 882 cases were recorded.

In November last year, Information and Broadcasting Ministry Secretary Apurva Chandra said that India has over 1.2 billion mobile phone users and 600 million smartphone users, making it a mature market. for spreading disinformation.

Indeed, Samir Saran, president of the Observer Research Foundation (ORF), tweeted that countries need to crack down on ChatGPT and other similar products.

“They need to undergo testing in multiple stages and various sandboxes before being made available – we are in uncharted territory here and the impact could be consequential,” he said on Wednesday.

Experts, however, have noted that since AI-driven models are generally built around the English language, most of the immediate impact will be on English-speaking countries. But it’s only a matter of time before templates are localized to suit different cultures, countries, nationalities and languages.

Read also | ETtech preview: Microsoft’s new AI-powered Bing chatbot, here’s what it looks like

Also, the models were built 12 months ago, so if someone asks about something happening in February 2023, the models won’t have an answer. So it’s not very current. Povolny added that ChatGPT will not remain as accessible and this could be a deterrent for bad actors.

“My main concern is with ChatGPT’s ability to make meaningful what previously wasn’t necessarily meaningful,” said Steve Povolny, principal engineer and head of advanced threat research at cybersecurity firm Trellix.

Global Concerns Spanish fact-checkers are preparing to deal with the potential onslaught of disinformation as the country approaches two elections: one national and one regional. Irene Larraz and her team work at one such Madrid-based company called Newtral.

“One potential risk we are bracing for is that people whose purpose is to distract fact checkers may use ChatGPT to spread the same disinformation written in different formats, so even though the input may be the same, the result is numerous articles that say the same thing in different ways and we can’t detect it in all platforms or in all places where it is widespread,” he explained.

Larraz said that just now there was an article published in a Spanish newspaper in which a reporter asked Bing linguistic processing if the president of Spain had a mustache and the tool said yes, but in reality Pedro Sánchez does not have a mustache. moustache.

“This is something small and silly that we may not care about, but while ChatGPT has the potential to invent answers to bigger problems that can result in polarization, it’s a big concern especially when dealing with sensitive topics,” he said.

Read also | ChatGPT, Bard & Ernie: The Three AI Musketeers

[ad_2]