The UK government has announced that it will host the first-ever global conference on AI safety this autumn (fall). The summit will see key countries, leading tech companies, and researchers come together to discuss safety measures to evaluate and monitor the most significant risks that AI poses.
“AI has an incredible potential to transform our lives for the better,” said Rishi Sunak, the UK’s Prime Minister. “But we need to make sure it is developed and used in a way that is safe and secure.”
“Time and time again throughout history we have invented paradigm-shifting new technologies and we have harnessed them for the good of humanity. That is what we must do again.”
“No one country can do this alone. This is going to take a global effort. But with our vast expertise and commitment to an open, democratic international system, the UK will stand together with our allies to lead the way.”
The risks from AI are multifaceted from the mundane to the science fiction. The most common danger from AI right now is that it can produce false information called hallucinations.
OpenAI is working on a method of processing in its LLMs that validates the logic at each stage to reduce the incidences of hallucinations but this model isn’t deployed for widespread use yet. In addition to this problem, AI can be used by malicious hackers to generate malware code that could be used in attacks.
While not really an AI safety issue, the situation around AI taking jobs is something else that politicians will have to contend with. Neowin reported last week, for example, on the fact that some companies have already started replacing people with generative AI.
Preceding the announcement of this summit, Rishi Sunak’s advisor on AI, Matt Clifford, said that AI could become very powerful within just two years if left unregulated, making it too difficult for humans to control, apparently. While the Terminator AI scenario feels like science fiction right now and probably will be for more than two years, it’s a good idea that governments look into the potential risks before it’s too late.