Artificial intelligence: ensuring security is difficult – 06/13/2023 – Market

Artificial intelligence experts generally follow one of the following schools of thought: artificial intelligence will greatly improve our lives or it will destroy us all. So how to ensure security?

Below, check out five of the challenges we face — and how the issue has been addressed in Europe.

1. Agree on what artificial intelligence is

The European parliament took two years to define an artificial intelligence system —software capable of, “for a given set of objectives defined by human beings, generating results such as content, predictions, recommendations or decisions that influence the environments with which they interact”.

This week, the European parliament is voting on the bill on the regulation of artificial intelligence – the first legal rules on the technology, which go beyond a voluntary code of conduct, requiring companies to comply with them.

2. Reaching a global consensus

Former head of the UK’s artificial intelligence office, Sana Kharaghani, notes that technology does not respect borders. “We need to have international collaboration on this – I know it’s going to be difficult,” she told BBC News. “This is not a domestic issue. These technologies don’t stay within one country’s borders.”

But there is still no plan for a United Nations-style global artificial intelligence regulatory body — although some have suggested that — and different territories have different ideas on the topic.

The European Union’s proposals are the strictest and include classifying artificial intelligence products depending on their impact — a spam filter for email, for example, would have lighter regulation than a cancer detection tool.

The UK, meanwhile, is handing over AI regulation to existing regulators — those who say the technology has discriminated against them, for example, are referred to the Equality Commission.

The United States, on the other hand, has only voluntary codes of conduct, and lawmakers recently admitted during an artificial intelligence committee hearing that they were concerned whether it was up to the task.

China intends to make companies notify users whenever an artificial intelligence algorithm is being used.

3. Ensuring public trust

“If people trust it, they will use it,” said IBM’s head of EU regulatory and government affairs Jean-Marc Leclerc. There are huge opportunities for artificial intelligence to improve people’s lives in incredible ways. Among them are:

• Help discover antibiotics
• Getting paralyzed people to walk again
• Addressing issues such as climate change and pandemics

But what about screening job applicants or predicting someone’s chance of committing a crime? The European Parliament wants the public to be informed about the risks associated with each artificial intelligence product.

Companies that break its rules can be fined up to 30 million euros (approximately R$ 157 million) – or 6% of global annual turnover.

But are developers able to predict or control how their product might be used?

4. Decide who writes the rules

Until now, artificial intelligence has largely been self-policing. The big companies say they comply with government regulation — “critical” to mitigating potential risks, according to Sam Altman, head of OpenAI, creator of ChatGPT.

But will they put profits above people if they get too involved in rule writing? You can bet they want to be as close as possible to the legislators in charge of setting the regulations.

And Martha Lane Fox, founder of Lastminute.com, says it’s important to listen not just to corporations. “We must involve civil society, academia, people who are affected by these different models and transformations”, she says.

5. Act fast

Microsoft, which has invested billions of dollars in ChatGPT, wants it to “take the drudgery out of the job”. It can generate text responses similarly to humans, but, Altman notes, it’s “a tool, not a creature.”

Chatbots should make professionals more productive. In some industries, artificial intelligence has the ability to create jobs and be a formidable assistant. In others, however, workers may lose their jobs—last month, BT (the British telecommunication company) announced that artificial intelligence would replace 10,000 jobs.

ChatGPT has been in public use for just over six months. Now he can write articles, plan vacations and pass professional exams. The capacity of these large-scale language models is growing at a phenomenal rate.

And two of the three “godfathers” of artificial intelligence — Geoffrey Hinton and Yoshua Bengio — are among those who warn that the technology has enormous potential for harm.

Europe’s Artificial Intelligence Act won’t come into effect until at least 2025 — “too late”, according to EU technology chief Margrethe Vestager.

She is formulating an interim voluntary code for the sector, together with the United States, which could be ready within weeks.