- Davis Cup organisers hit back at critics of Nadal retirement ceremony
- Noel in a 'league of his own' as he wins Gurgl slalom
- A dip or deeper decline? Guardiola seeks response to Man City slump
- Germany goes nuts for viral pistachio chocolate
- EU urges immediate halt to Israel-Hezbollah war
- Far right targets breakthrough in Romania presidential vote
- Basel votes to stump up bucks to host Eurovision
- Ukraine shows fragments of new Russian missile after 'Oreshnik' strike
- IPL auction records tumble as Pant and Iyer snapped up
- Six face trial in Paris for blackmailing Paul Pogba
- Olympic champion An wins China crown in style
- It's party time for Las Vegas victor Russell on 'dream weekend'
- Former Masters champion Reed seals dominant Hong Kong Open win
- Norris applauds 'deserved' champion Verstappen
- Jaiswal and Kohli slam centuries as Australia stare at defeat
- Kohli blasts century as India declare against Australia
- Verstappen 'never thought' he'd win four world titles
- Former Masters champion Reed wins Hong Kong Open
- Awesome foursomes: Formula One's exclusive club of four-time world champions
- Smylie beats 'idol' Cameron Smith to win Australian PGA Championship
- Five key races in Max Verstappen's 2024 title season
- Max Verstappen: Young, gifted and single-minded four-time F1 champion
- 'Star is born': From homeless to Test hero for India's Jaiswal
- Verstappen wins fourth consecutive Formula One world title
- Survivors, sniffing dogs join anti-mine march at Cambodia's Angkor Wat
- Far right eye breakthrough in Romania presidential vote
- Jaiswal slams majestic 161 but Australia fight back in Perth
- Edinburgh's alternative tour guides show 'more real' side of city
- IPL teams set to splash the cash at 'mega-auction' in Saudi Arabia
- Olympics in India a 'dream' facing many hurdles
- Wounded Bangladesh protesters receive robotic helping hand
- Majestic Jaiswal 141 not out as India pile pain on Australia
- Giannis, Lillard lead Bucks over Hornets as Spurs beat Warriors
- Juan Mata agent slammed as 'cowardly' by angry A-League coach
- Marta inspires Orlando Pride to NWSL title
- Palestinian pottery sees revival in war-ravaged Gaza
- Main points of the $300 billion climate deal
- Robertson wants policy change for overseas-based All Blacks
- Israel retreat helps rescuers heal from October 7 attack
- Afghan women turn to entrepreneurship under Taliban
- Mounting economic costs of India's killer smog
- At climate talks, painstaking diplomacy and then anger
- Uruguayans head to polls with left hoping for comeback
- Trump's mass deportation plan could end up hurting economic growth
- Iran director in exile says 'bittersweet' to rep Germany at Oscars
- US consumers to bargain hunt in annual 'Black Friday' spree
- Cheers, angst as US nuclear plant Three Mile Island to reopen
- Scientists seek miracle pill to stop methane cow burps
- Australia ditches plans to fine tech giants for misinformation
- Developing nations slam 'paltry' $300 bn climate deal
Inbred, gibberish or just MAD? Warnings rise about AI models
When academic Jathan Sadowski reached for an analogy last year to describe how AI programs decay, he landed on the term "Habsburg AI".
The Habsburgs were one of Europe's most powerful royal houses, but entire sections of their family line collapsed after centuries of inbreeding.
Recent studies have shown how AI programs underpinning products like ChatGPT go through a similar collapse when they are repeatedly fed their own data.
"I think the term Habsburg AI has aged very well," Sadowski told AFP, saying his coinage had "only become more relevant for how we think about AI systems".
The ultimate concern is that AI-generated content could take over the web, which could in turn render chatbots and image generators useless and throw a trillion-dollar industry into a tailspin.
But other experts argue that the problem is overstated, or can be fixed.
And many companies are enthusiastic about using what they call synthetic data to train AI programs. This artificially generated data is used to augment or replace real-world data. It is cheaper than human-created content but more predictable.
"The open question for researchers and companies building AI systems is: how much synthetic data is too much," said Sadowski, lecturer in emerging technologies at Australia's Monash University.
- 'Mad cow disease' -
Training AI programs, known in the industry as large language models (LLMs), involves scraping vast quantities of text or images from the internet.
This information is broken into trillions of tiny machine-readable chunks, known as tokens.
When asked a question, a program like ChatGPT selects and assembles tokens in a way that its training data tells it is the most likely sequence to fit with the query.
But even the best AI tools generate falsehoods and nonsense, and critics have long expressed concern about what would happen if a model was fed on its own outputs.
In late July, a paper in the journal Nature titled "AI models collapse when trained on recursively generated data" proved a lightning rod for discussion.
The authors described how models quickly discarded rarer elements in their original dataset and, as Nature reported, outputs degenerated into "gibberish".
A week later, researchers from Rice and Stanford universities published a paper titled "Self-consuming generative models go MAD" that reached a similar conclusion.
They tested image-generating AI programs and showed that outputs become more generic and strafed with undesirable elements as they added AI-generated data to the underlying model.
They labelled model collapse "Model Autophagy Disorder" (MAD) and compared it to mad cow disease, a fatal illness caused by feeding the remnants of dead cows to other cows.
- 'Doomsday scenario' -
These researchers worry that AI-generated text, images and video are clearing the web of usable human-made data.
"One doomsday scenario is that if left uncontrolled for many generations, MAD could poison the data quality and diversity of the entire internet," one of the Rice University authors, Richard Baraniuk, said in a statement.
However, industry figures are unfazed.
Anthropic and Hugging Face, two leaders in the field who pride themselves on taking an ethical approach to the technology, both told AFP they used AI-generated data to fine-tune or filter their datasets.
Anton Lozhkov, machine learning engineer at Hugging Face, said the Nature paper gave an interesting theoretical perspective but its disaster scenario was not realistic.
"Training on multiple rounds of synthetic data is simply not done in reality," he said.
However, he said researchers were just as frustrated as everyone else with the state of the internet.
"A large part of the internet is trash," he said, adding that Hugging Face already made huge efforts to clean data -- sometimes jettisoning as much as 90 percent.
He hoped that web users would help clear up the internet by simply not engaging with generated content.
"I strongly believe that humans will see the effects and catch generated data way before models will," he said.
A.Gasser--BTB