Page 65 - CBM26

~ Spotlight On ~

to force them to do so. details any data used from publications, songs, etc. A copyright
holder can use that registry to pursue an alleged infringement.
“We must ask ourselves: should we develop nonhuman minds However, it does not pose an opt-in requirement like the GDPR
that might eventually outnumber, outsmart, obsolete and replace regime that some copyright groups are demanding.
us?” the letter stated. “Should we risk the loss of control of our
civilisation?” Such decisions must not be delegated to unelected While the European Parliament proposal still faces a vote in the
tech leaders.” More than three years ago, European Union General Assembly, tentatively in June or July, it is expected to sail
lawmakers in Brussels took up the societal challenges outlined by through after a near consensus was achieved at the committee
FLI (and others) and began work on the world’s first regulatory level. Negotiations will begin soon with the Council of Ministers
artificial intelligence framework that it hopes will mesh with to iron out differences such as those on law enforcement AI
its data privacy regime known as GDPR. The legislation, which surveillance. “Against conservative wishes for more surveillance
categorises AI applications into four risk groups and calls for a and leftist fantasies of over-regulation, the European
ban on those considered most harmful, is expected to be finalised Parliament found a solid compromise that would regulate
by the end of 2023. AI proportionately, protect citizen’s rights as well as foster
innovation and boost the economy,” Svenja Hahn, a German
When the EU legislative work began, the highest priority MEP and member of the Renew Europe political group, said
was the protection of human rights from nefarious AI-driven after the committee vote.
applications such as social scoring in labour markets or police
surveillance using facial and emotional recognition, biometrics, While the AI regulatory push moves forward, including
and crime prediction. The Orwellian reality unfolding in China, pending legislation in the United States, the race to capitalise
where AI technologies have been employed to help smother on AI-powered chatbots and other applications has taken off.
democratic dissent in Hong Kong and other parts of the country, The leaders, of course, are the Big Tech companies as they have
was considered a wake-up call. Three years on, there is an EU the massive data sets needed to effectively feed the AI neural
consensus to ban social scoring from preventing labour market networks. Microsoft, which helped fund Open AI ChatGPT,
discrimination. But there are divisions over the extent AI is now using it for its search engine Bing. With speculation
surveillance applications such as facial recognition (FR) should rife that Google’s search engine dominance was under threat,
be allowed as a law enforcement tool. EU member states insist it unveiled Bard in February. Its share market value promptly
it is vital, especially for combatting terrorism. The role that facial dropped by $100 billion in a day the chatbot erroneously
recognition has played in helping U.S. law enforcement track answered questions. But since that inauspicious Bard launch,
down rioters who violently stormed the U.S. Capitol on Jan. 6, Google has more than recovered. It announced on May 10 a new
2021, to support Donald Trump’s coup d’etat is considered a AI-driven search engine called Search Generative Experience.
relevant example of how AI-driven FR is a necessary tool. Investors have ploughed billions of dollars into it, as well as
Microsoft and others, including Nvidia, whose advanced chip
The European Parliament, which tentatively finalised on May 11 technology will drive AI generative applications.
its version of the AI legal framework, wants a law enforcement
ban on live AI FR in public spaces, including nondiscriminatory As the stock market bets that chatbots are a smartphone-like
scraping of facial images as well as predictive policing, biometric paradigm shift paved with gold, there are still sceptics. Does
categorisation, and emotional recognition. Meanwhile, anybody remember Watson? Its creator, IBM, certainly does.
“generative AI” is a good example of how rapidly AI technology When its AI program defeated all challengers in the American
has evolved in a brief time and how the multiple societal impact quiz TV game show Jeopardy in 2010, the achievement was also
concerns have grown exponentially in recent months. When the hailed as transformative and would restore the legacy computer
European Commission launched the EU legislative proposal maker to its glory days when it dominated the mainframe
in 2021, it barely addressed the issue. It is also absent from computer market in the 1960s and 70s.
the legislative version approved by the EU member states in
December. However, IBM’s shoot-for-the-moon revolutionary plans for
Watson flopped. In recent years, IBM scaled back its Watson
But when the Open AI ChatGPT-3 hit the market in ambitions and has found niche markets in the medical, education
November, followed up by the ChatGPT-4 in March, and and business world. As AI technology advances, it remains
deepfakes flourished, the European Parliament lawmakers unclear if the lessons learned from Watson will pave the way
promptly drew up a generative AI regime they believe will to the future. Or is HAL lurking in the shadows, ready to seek
protect copyright holders. It includes requirements that chatbot revenge? ••
developers provide an algorithm documentation registry that

Alan Rhode is a co-founder at Taxmen, the one-stop-shop for legal and tax services to the e-commerce industry.
Joe Kirwin is a US-born journalist based in Brussels.

65