Public Health Experts Sound Alarm on AI's Potential Misuse

John Lopez, Tech Times 10 May 2023, 08:05 am

AI has already been heralded as a game changer in healthcare. In CES 2023, for instance, experts from the American Medical Association said that artificial intelligence could soon take over the domain of medicine, but in a way that benefits both physicians and patients.

But a group of doctors and public health experts have now cautioned that it could pose an existential threat to humanity.

The group, which included experts from the International Physicians for the Prevention of Nuclear War and the International Institute for Global Health, released an article in BMJ Global Health pushing for better AI regulation.

Doctors Pushing for AI Regulation

The article focuses on the hazards of AI in healthcare, such as errors that could hurt patients, data privacy and security concerns, and the potential for AI to exacerbate social and health inequalities by incorporating existing biases into automated algorithms.

OpenAI's ChatGPT, for example, has already been chastised for its data privacy laws, which have been the focus of contention in Italy, which has recently banned and reinstated the service.

Additionally, biases are known to exist in OpenAI's image-generating tool DALL-E 2, and other AI models can have, produce, and perpetuate biases.

The authors urge the medical and public health communities to learn more about the possible concerns posed by AI and to participate in policy debates about reducing these risks while reaping the advantages of AI.

Exploring Other Threats Posed by AI

AI's potential for harm is not restricted to healthcare. The report emphasizes the hazards of employing AI for surveillance and information campaigns, which might erode democracy further by generating a widespread breakdown in trust or fueling societal division and violence, with resulting public health consequences.

One example is the rapid development of emotion analysis technology, which the UK's information rights watchdog believes may endanger vulnerable people.

In 2023, the agency plans to issue recommendations on using biometric technologies like voice, facial, and fingerprint recognition.

The authors also expressed concern about the development of future weapons systems that might find, choose, and kill "at an industrial scale" without the need for human oversight.

Furthermore, the research discusses AI's possible impact on jobs, noting that unemployment is highly related to poor health outcomes and behavior.

While there would be numerous benefits to eliminating repetitive, risky, and unpleasant employment, the authors warn that authorities must confront the possibility of job losses due to AI-driven automation.

Reproducibility Crisis

The health industry has been wrestling with the potential benefits of AI, such as enhanced disease diagnosis, discovery of new medicines, answering patient concerns, and performing menial jobs.

However, Axios tells us that research has found that hospital algorithms discriminate against Black patients by allocating less attention to them. Algorithm dependability has also been questioned, with experts warning of a "reproducibility crisis" in healthcare studies.

The paper's authors are not the only ones raising concerns about AI. Geoffrey Hinton dubbed the "godfather of AI," announced his departure from Google due to concerns about machine learning dangers.

The United Nations and the European Union have taken moves to regulate AI, but more has to be done.

Stay posted here at Tech Times.