[ad_1]
As Silicon Valley races to build powerful and popular artificial intelligence systems, troubling “incidents” ranging from convincing AI deepfakes, banking fraud, bias and even deaths will increase this year, a tech expert says.
Following the release of ChatGPT last November, tech companies have been rushing to develop powerful AI systems to keep the pace with competitors.
The AI Incident Database, which is run by nonprofit Responsible AI Collaborative, tracks various incidents caused by AI and is projected to record double the number of incidents this year compared to last. The database defines incidents through examples such as an autonomous car killing a pedestrian, a “trading algorithm” causing a “market ‘flash crash’ where billions of dollars transfer between parties,” or a “facial recognition system” causing “an innocent person to be arrested.”
“There is no such thing as a minor incident with AI,” said Sean McGregor, the founder of the AI Incident Database who previously worked as a machine learning architect.
“If I produce a system that makes 8 billion people feel slightly more depressed, then that system will have pushed some number of people to suicide,” McGregor told Fox News Digital. “The unfortunate matter is that with fantastical scale even small impacts add up.”
REGULATORS SHOULD KEEP THEIR HANDS OFF AI AND FORGET MUSK-BACKED PAUSE: ECONOMIST
Following the release of ChatGPT last November, tech companies have been rushing to develop powerful AI systems to keep the pace with competitors. (Gabby Jones / Bloomberg via Getty Images / File)
The AI Incident Database tracks the “history of harms or near harms” caused by the proliferation of AI technology in order “to learn from experience so we can prevent or mitigate bad outcomes,” according to the database’s website. McGregor said the website is a response to philosopher George Santayana’s aphorism, “Those who cannot remember the past are condemned to repeat it.”
“We must remember past incidents so we can engineer a better future,” McGregor said.
The tech expert believes that incidents caused by AI will likely skyrocket this year, according to Newsweek. Overall, the database has found more than 500 examples of AI incidents from 2003 to this year. Last year ended with 90 incidents, and there have already been 45 incidents this year between January and March.
CHATGPT AI LISTS JOBS IT CAN DO BETTER THAN HUMANS AS MILLIONS COULD BE PUT OUT OF WORK
“We expect AI incidents to far more than double in 2023 and are preparing for a world where AI incidents are likely to follow some version of Moore’s Law,” McGregor told Newsweek, which projected 2023 will see about 180 incidents, based on the data’s current rate of incidents.
Moore’s Law is an observation from an Intel co-founder that transistors on a circuit would double every two years, making the speed and capability of computers also increase.
As Silicon Valley races to build powerful and popular artificial intelligence systems, troubling “incidents” ranging from convincing AI deepfakes, banking fraud, bias and even deaths will increase this year, a tech expert says. (Reuters / Dado Ruvic / Illustration / File)
The database – which allows users to search by information such as the year of the incident, the developer behind the AI system or a description of the incident – shows that 20 years ago, there were four reports of AI incidents. The incidents include a claim that a maternity ad from Target allegedly predicted a teenager was pregnant before her own family by using an algorithm – though the claim has been called into question.
It also includes two friendly-fire military incidents that left both British and American service members killed during the Iraq War.
AI: NEWS OUTLET ADDS COMPUTER-GENERATED BROADCASTER ‘FEDHA’ TO ITS TEAM
The database shows that various incidents have increased since, notably jumping from 37 incidents in 2019 to 78 in 2020, before notching what is currently its highest number of incidents in 2022 at 90.
“Airplanes can’t be designed to survive a head-on crash, but that is why we have air traffic controllers,” McGregor told Fox News Digital. “Engineering safety requires as many advancements in culture as it requires advances in technology and that is what the AI Incident Database is working to support.”
McGregor said he hopes the database will “also come to dispel the notion that ChatGPT and similar systems are somehow god machines that are increasingly going rogue.” (Gabby Jones / Bloomberg via Getty Images / File)
He added that he hopes the database will “also come to dispel the notion that ChatGPT and similar systems are somehow god machines that are increasingly going rogue.”
“They are behaving as we would expect them to behave based on how they are engineered,” he said.
GOOGLE SCRAMBLES FOR NEW SEARCH ENGINE AS AI CREEPS IN: REPORT
This year alone, the database reported on incidents such as a Belgian father allegedly committing suicide after talking with a chatbot for weeks about the climate, phony images of former President Donald Trump getting arrested, and Pope Francis wearing a ritzy white puffer jacket.
“The biggest AI incident threat we are facing right now is complacency. Through each incident we may become acculturated to the risks of intelligent systems rather than adopt a ‘never again’ viewpoint,” McGregor said.
The website of Midjourney, an artificial intelligence capable of creating AI art, is seen on a smartphone on April 3, 2023, in Berlin. (Thomas Trutschel / Photothek via Getty Images / File)
The launch of ChatGPT in November served as a watershed moment that reverberated across the tech industry and catapulted various companies to compete on building their own AI systems.
OpenAI’s ChatGPT became the fastest-growing user base with 100 million monthly active users in January as people across the world rushed to use the chatbot, which simulates human-like conversations based on prompts it is given.
Last month, thousands of tech leaders, experts and others signed an open letter calling on all AI labs working on tech more powerful than ChatGPT to pause for at least six months to roll out safety regulations.
Twitter and Tesla CEO Elon Musk and Apple co-founder Steve Wozniak were among the signatories who warned that such computer intelligence “can pose profound risks to society and humanity.”
CLICK HERE TO GET THE FOX NEWS APP
Despite the calls to pause, the AI race to create the most powerful system is in full swing, with Google working to overhaul its search engine and even create a new one that relies on AI, while Microsoft has rolled out the “new Bing” search engine described as users’ “AI-powered copilot for the web.”
Source link
