Should Artificial Intelligence be limited? This question is a hotly debated one in artificial intelligence circles. As AI powers our world, it is crucial to balance its negative effects with its positive ones. In the recent AlphaFold 2 breakthrough, AI-powered systems have become more powerful and sophisticated, but the associated dire warnings have hardly been addressed. So, should we limit AI to only certain tasks? And should we also limit AI to the research arena?
If we overestimate AI’s potential, we may risk damaging the advancement of science. For example, some argue that digitization can help us understand the past. For their project, they analyzed five million English books published since 1800. Compared to the past, these works had more English words. These new technologies could also lead to technological unemployment in the future. Therefore, limiting AI’s use to certain fields may be the best way to avoid overestimating the impact on society.
However, limiting AI can have adverse effects and should not be governed by a single authority. AI-based solutions have the potential to change the way we live in our daily lives. They can do everything from detecting cancer to reducing airplane collisions. The application of AI programs is so wide-ranging that it could even be implemented in nonautonomous vehicles. While there are a few controversial applications of AI, this is not the only use for AI.
While there is no single best AI, narrow AI is a powerful tool that can perform superhuman tasks and demonstrate superior creativity. In 2012, a self-driving Toyota Prius completed ten hundred-mile journeys on its own, setting the path to driverless cars. Meanwhile, in 2011, an AI program called IBM Watson won the US quiz show Jeopardy!, using natural language processing and analytics to process vast data repositories and answer questions in fractions of a second.
AI is already a threat to humankind, and some fear a hypercapable AI system will take over the world and kill humanity. But should we limit AI? Despite widespread fear, there are several reasons to limit AI. Some people fear that it will end the world as we know it and others fear that the technology will be misused. Others are convinced that AI must be regulated. Some famous figures, like Bill Gates and Mark Zuckerberg, have been adamantly opposed to regulation.
While it is unlikely that AI will make the wrong decisions, it may improve the odds of fairness and stakeholder trust. However, it comes at a price. And it may not be worth it at all times. In the end, users of AI must decide whether the costs of unfair outcomes are worth the benefits of more accurate output. In the end, it is better to limit AI than to allow it to take over the world.
Some debates about legal liability for AI systems have centered on legal liability. While algorithm operators likely fall under product liability rules, the legal status of AI operators depends on the specific facts and circumstances of the case. The penalty could range from a fine to a prison sentence for major harms. The Uber fatality will be an important test case for AI’s liability, especially since the state actively recruited Uber to test autonomous vehicles. In addition to that, it granted Uber broad latitude for its road testing.
There are concerns about the ethical implications of AI’s decision-making abilities. There are already laws to limit AI, but they still do little to keep criminals from exploiting it. The dangers of AI are too high to ignore. Further, there is a general lack of awareness about the implications of AI’s impact on society. If these technologies are not regulated, people may be hesitant to use them for their own ends.
There is a need to regulate AI, but it is necessary to limit the use of AI in ways that do not impede fundamental human rights. Using transparency is an interim solution. But transparency must be accompanied by robust regulation. While transparency may be a long-term solution, it may not be enough. For example, we should ensure that AI is not used for discriminatory purposes, and only when it threatens our societies and fundamental rights.
Current AI can’t cope with the unforeseen changes in goals and circumstances. Even when it does, it can’t reason from a general perspective. That’s why AI must be limited to narrow areas. However, if we’re successful, we might be able to create an AGI that is superior to humans in every area. We may consider this a holy grail. In the meantime, AI should be limited to a few narrow fields, and we need to be realistic about the limits.