CheckMate is a weekly newsletter from RMIT FactLab recapping the latest in the world of fact checking and misinformation. It draws on the work of FactLab’s researchers and journalists, including its CrossCheck unit, and of its sister organisation, RMIT ABC Fact Check.
You can subscribe to have the next edition delivered straight to your inbox.
CheckMate April 28, 2023
This week, we ask the experts about supposedly “explosive” new research on face masks and health risks.
We also bring you a fact check on a claim about the gender pay gap from Opposition Leader Peter Dutton, and look into the murky world of AI-generated music.
Experts weigh in on ‘explosive’ mask study
Experts have expressed doubt over a new research paper purportedly showing that face masks “may raise the risk of stillbirths, testicular dysfunction and cognitive decline in children”.
The paper, authored by German researchers and featured in a sensational Daily Mail article that described the findings as “explosive”, reviewed numerous studies regarding carbon dioxide (CO2) exposure and face masks.
According to University of Wollongong epidemiologist Gideon Meyerowitz-Katz, it then “used some rodent research to speculate on how this [exposure] might cause issues to long-term health”.
The Daily Mail article was shared widely online, including by former federal MP and current United Australia Party national director Craig Kelly.
But Mr Meyerowitz-Katz, along with other experts, told CheckMate the study did not actually find any link between mask wearing and negative health outcomes.
“The study itself is a scoping review, and … does not prove much about masks,” he said.
“At best, this might be useful to drive future research, but it’s not the sort of methodology that can prove a causal connection, especially as none of the included studies actually measured the long-term health of people wearing masks.”
James Trauer, who heads the epidemiological modelling unit at Monash University’s School of Public Health and Preventive Medicine, said the methods used in the review were “not really adequate to the importance of the question” being investigated and didn’t “justify the conclusions [the authors] arrived at”.
As the paper was not a systematic review that followed established guidelines, he said it was “hard to know exactly what they’ve done”, but noted that the absence of any studies contradicting its main findings was “highly suspicious”.
Running through those findings, Dr Trauer told CheckMate he was “happy to accept” that masks increase the amount of carbon dioxide inhaled by wearers and that — based on the studies listed — they “probably” increase the concentration of CO2 in humans.
It was, however, “very dubious” to conclude from the evidence presented that increased CO2 in animals leads to negative health effects, he said, adding that the paper contained “no evidence whatsoever” that increased CO2 in the blood leads to health problems in humans.
Indeed, given that scoping reviews aim to identify what evidence is out there, Dr Trauer said that “possibly the main conclusion … should be that they have not been able to identify any evidence of any health effects in humans from increased carbon dioxide”.
When it came to the specific suggestion from the authors that mask wearing “may be related to current observations of stillbirths”, Shaun Brennecke, director of the Department of Maternal-fetal Medicine at the Royal Women’s Hospital in Melbourne, described the evidence as “very circumstantial” and “not in any way shape or form conclusive”.
While there had indeed been a pandemic-era rise in stillbirths, he said, there was “no causative study amongst any of it to actually directly link [it with] face mask wearing”.
Crucially, Professor Brennecke said that an Australian study referenced by the paper showed a rise in preterm stillbirths only, and not in stillbirths among term pregnancies.
“If the face mask theory was likely to be significant,” he said, “you would have expected [that] the women who reached term would also have had an increase in stillbirths, because they’ve actually been wearing masks longer than the women who were preterm.”
The authors of the German paper have also come under scrutiny from US-based fact checkers at FactCheck.org this week for a different paper that suggested long COVID symptoms may actually be the result of mask wearing.
Contrary to the paper’s assertion, however, the fact checkers found there was “no evidence that face masks, when used as recommended, have a negative impact on the respiratory system that could lead to any serious health problems”.
Finally, the Daily Mail article prominently features a link to another story which alleges that “wearing a mask can expose children to dangerous levels of carbon dioxide in just three minutes”.
That story, from July 2021, was based on a research letter that was retracted by the publishing journal within a month (something made clear by the stamping of a “RETRACTED” watermark repeatedly across the original letter).
Do algorithms dream of R&B?
Content generated by artificial intelligence made headlines again this week after a song featuring vocal tracks — sort of — from two of the world’s biggest R&B stars took off on social media and streaming apps.
Uploaded to YouTube and Spotify by a user called “Ghostwriter” and clocking up more than 15 million views on TikTok, the song “Heart on my sleeve” used generative AI to mimic the voices of artists The Weeknd and Drake in a collaboration to which neither consented.
While certainly not the first case of an artist’s voice being plagiarised by the new technology (one recent example being an AI “Kanye West” covering the Justin Bieber hit Love Yourself), The New York Times labelled the Drake controversy “the latest and loudest example of a grey-area genre that has exploded in recent months” — namely, the creation of homemade tracks that use AI “to conjure familiar sounds that can be passed off as authentic, or at least close enough”.
“It earned instant comparisons to earlier technologies that disrupted the music industry, including the dawn of the synthesiser, the sampler and the file-sharing service Napster,” the article noted.
Universal Music Group, the label representing both artists, took a dim view, however, as it swooped in to have the tracks removed from platforms.
In a statement, Universal said using its artists’ music to train AI algorithms not only breached copyright law but also foreshadowed a reckoning in the music industry in which people would need to choose “which side of history” they were on: “the side of artists, fans and human creative expression, or … the side of deep fakes, fraud and denying artists their due compensation”.
Looking beyond the art world, AI also made its first major foray into the US presidential election, with the Republican Party incorporating a series of AI-generated images into a new 2024 campaign video attacking President Joe Biden.
Alex Thompson, national political correspondent for the US news website Axios, wrote in a tweet that the video was the party’s first to use 100 per cent AI-generated images.
It attempts to paint a dystopian picture of how the future might look if “the weakest president we’ve ever had were re-elected”, suggesting it could include war over Taiwan, collapsing financial systems and escalating crime.
Why there’s more to the story with Peter Dutton’s gender pay gap claim
Opposition Leader Peter Dutton has been busily spruiking a long list of plans for the future as well as the Coalition’s recent achievements, including its record on women’s economic issues.
“Under Labor, the gender pay gap was at 17.4 per cent and reduced to 13.8 per cent on our watch,” Mr Dutton told the national summit of the Council of Small Business Organisations of Australia on April 4.
But RMIT ABC Fact Check this week found there’s more to the story than Mr Dutton’s claim suggests.
Indeed, the gender pay gap was at 17.4 per cent when Labor left office in late 2013.
The pay gap also fell steadily during most of the Coalition’s time in government, reaching 13.8 per cent in November 2021 (although it did rise again in the months prior to the May 2022 election and stood at 14.1 per cent when the Coalition was ousted).
That said, the gap also peaked during the Coalition’s nine-year reign at 18.7 per cent, which was higher than at any time during the Rudd-Gillard or Howard governments.
Moreover, since the election of the Albanese government, the pay gap has fallen to a historical low of 13.3 per cent.
Experts told Fact Check that changes in the gender pay gap were mostly driven by wider economic conditions, but government policy could also play its part in addressing the issue.
Edited by Ellen McCutchan and David Campbell
Got a fact that needs checking? Tweet us @ABCFactCheck or send us an email at factcheck@rmit.edu.au