Experts in artificial intelligence typically subscribe to one of two schools of thought: either it will significantly improve our lives or it will obliterate humanity as we know it. The discussion about technology regulation in the European Parliament this week is crucial for this reason. Here are five obstacles to overcome in order to make AI safe.
A definition of an AI system has taken the European Parliament two years to develop. It states that an AI system is "software that, for a given set of human-defined objectives, generates outputs such as content, predictions, recommendations, or decisions influencing the environments they interact with.".
It is voting this week on its Artificial Intelligence Act, which goes beyond voluntary codes and mandates that businesses comply. This is the first set of laws of its kind on AI. .
Technology disregards national boundaries, according to Sana Kharaghani, a former director of the UK Office for Artificial Intelligence.
She tells BBC News, "I know it will be hard, but we do need international collaboration on this.". It's not a domestic issue. These technological advancements transcend national borders.
However, despite some suggestions to the contrary, there is still no plan for a global United Nations-style AI regulator, and various regions have their own ideas.
- The strictest proposals come from the European Union and include classifying AI products based on their effects; a spam filter for email, for example, would be subject to less regulation than a tool for detecting cancer.
- The United Kingdom is integrating existing regulatory bodies for AI regulation; for example, those who claim the technology has discriminated against them would go to the Equalities Commission.
- In the United States, there are no mandatory codes; lawmakers acknowledged in a recent hearing of the AI committee that they had doubts about their suitability for the job.
- China wants to mandate that businesses notify customers any time an AI algorithm is being used.
International Business Machines (IBM) Corporation EU government and regulatory affairs head Jean-Marc Leclerc claims that if people trust it, they will use it.
There are numerous possibilities for AI to significantly enhance people's lives. There is already:.
- aiding in the discovery of antibiotics.
- walking again after being paralyzed.
- dealing with concerns like pandemics and climate change.
But what about selecting candidates for jobs or determining someone's propensity for crime?
The European Parliament wants the public to be aware of the risks associated with each artificial intelligence product.
Companies that violate its rules risk being fined up to €30 million or 6% of their annual global revenue, whichever is greater.
Can designers, however, foresee or control how their product might be used?
AI has largely self-regulated up until this point.
Sam Altman, CEO of OpenAI, the company that developed ChatGPT, claims that the major corporations support government regulation because it is "critical" to reduce the risks.
But if they get too involved in creating the rules, will they prioritize profits over people?
They want to be as close as possible to the legislators who are responsible for establishing the rules, you can bet on that.
Additionally, Baroness Lane-Fox, the founder of Lastminute.com, asserts that it's important to hear other voices besides big business.
She asserts that we need to involve the academic community, civil society, and those who are impacted by the various models and transformations.
Microsoft hopes ChatGPT will "take the drudgery out of work" thanks to its massive investment in it.
Although it is "a tool, not a creature," Mr. Altman notes that it can produce text responses and prose that sounds like human speech.
Worker productivity is supposed to increase with chatbots.
Additionally, AI has the potential to generate jobs and serve as a powerful assistant in some industries.
However, other people have already lost them. Last month, BT declared that AI would eliminate 10,000 jobs. .
Just over six months have passed since ChatGPT was made available to the general public.
Now, it is capable of planning vacations, writing essays, and passing professional exams.
These massive language models are becoming increasingly more capable.
Geoffrey Hinton and Prof. Yoshua Bengio, two of the three "godfathers" of artificial intelligence, have been among those to warn that the technology has a great potential for harm.
The Artificial Intelligence Act won't take effect until at least 2025, which is "way too late," according to Margrethe Vestager, the EU's technology commissioner.
Together with the US, she is creating an interim voluntary code for the industry that could be completed in a matter of weeks.
