The European Union (EU) took a major step on Wednesday in giving the world its first set of governance rules surrounding artificial intelligence (AI), as the European Parliament passed a draft law known as the A.I. Act.
The A.I. Act, which was first proposed by the EU on April 21, 2021, has been touted as the world’s first and most comprehensive regulatory framework. Since its proposal, the European Commission, the Council of the European Union, and European Parliament have been working on modifying and refining its initial draft, with a final version not expected until later this year.
While the initial draft came prior to the surge of generative AI, including chatbots, this new draft law certainly takes into consideration generative AI systems like OpenAI’s ChatGPT and the implications they bring.
Testing Before Deployment
The Act, which applies to all EU member nations, requires EU member states to establish at least one regulatory “sandbox” to test AI systems before they are deployed.
“The one thing that we wanted to achieve with this text is balance,” Dragoș Tudorache, a member of the European Parliament, told journalists. The Act protects citizens while also “promoting innovation, not hindering creativity, and deployment and development of AI in Europe,” he added.
Greatest Potential for Human Harm
At the heart of Wednesday’s passed bill, generative AI would be subject to new transparency requirements, which include publishing summaries of copyrighted material used for training the system as well as implementing safeguards to prevent the AI from generating illegal content.
The bill’s “risk-based” approach to regulating AI focuses on use cases with the greatest potential for human harm, including the utilization of AI systems to operate critical infrastructures like water or energy, our legal justice system, and determining access to public services and government benefits.
Developers of this technology would be required to conduct risk assessments before putting that technology into the mainstream, similar to how the drug approval process currently operates.
Banning of Facial Recognition
Another heated area of debate revolves around the use of facial recognition technology. The European Parliament voted to ban the use of live facial recognition, but is still open to whether there should be an exception for its use in cases of national security and other law enforcement purposes.
Adding to that, another provision of the A.I. Act would ban companies from scraping biometric data from social media to help build out databases – something that was a major issue after facial-recognition company Clearview AI was fined $20 million USD by France’s data protection authority (CNIL) for the illegal collection and processing of biometric data belonging to French citizens, on top of an overdue $5.2 million USD fine.
Generative AI to Add $4.4 Trillion USD to Global Economy?
Generative AI is estimated to add $4.4 trillion USD to the global economy, according to a Wednesday report published by McKinsey & Company.
While that figure is at the upper-end of McKinsey’s projected range for generative AI’s value, the report titled “The Economic Potential of Generative AI” also provides the lower end of the range which rests at $2.6 trillion USD.
During the Viva Conference in Paris on Wednesday, Meta’s chief AI scientist Yann LeCun spoke to the current limitations of generative AI, specifically those that are trained on large language models.
According to LeCun, generative AI that is solely trained on language aren’t very intelligent.
“Those systems are very limited, they don’t have any understanding of the underlying reality of the real world, because they are purely trained on text, massive amounts of text,” LeCun said. “Most of human knowledge has nothing to do with language…so that part of the human experience is not captured by AI.”
“We have made history today,” Brando Benifei, a member of the European Parliament working on the EU AI Act, told journalists.
“While Big Tech companies are sounding the alarm over their own creations, Europe has gone ahead and proposed a concrete response to the risks AI is starting to pose,” Benifei added.
This week, more than 40% of business leaders at the Yale CEO Summit, including Walmart chief Doug McMillion and Coca-Cola CEO James Quincy, all shared their belief that AI has the potential of destroying humanity five to 10 years from now.
The AI Act takes the penalties that we’ve seen from Europe’s GDPR privacy framework a step further, where companies engaged in prohibited AI practices could be fined up to $43 million USD or an amount equal to 7% of a company’s worldwide annual turnover, whichever is higher.
GDPR currently structures its fines up to $10.8 million USD, or up to 2% of a firm’s global turnover, whichever is higher.
In other news, read about hyperdimensional computing and why we may be looking at AI all wrong.
Click here to view full gallery at Hypemoon