'Malicious' predictive text bot released after AI code of ethics published
An artificial intelligence (AI) program deemed too malicious to release has gone public, as the Australian government issued a series of principles of the new technology.
GPT-2 takes predictive text to the next level, allowing the bot to create long tracts of writing on its own, practically indistinguishable from humans.
Developed by Elon Musk's company OpenAI, GPT-2 could be used to convincingly scam people or create "synthetic propaganda" for terror or extremist groups.
But they released the program overnight after seeing "no strong evidence of misuse so far".
In Australia, the eight voluntary principles issued by the Department of Industry, Innovation and Science are intended to reduce the risk of negative impacts of the development of AI.
AI should be programmed to benefit individuals, society and the environment, respect human rights, diversity and autonomy and be inclusive and accessible.
AI systems should also respect and uphold privacy rights and data protection.
"We need to make sure we're working with the business community as AI becomes more prevalent and these principles encourage organisations to strive for the best outcomes for Australia and to practice the highest standards of ethical business," science minister Karen Andrews said.
"This is essential, as we build Australians' trust that AI systems are safe, secure, reliable and will have a positive effect on their lives."
Requiring artificial intelligence to respect diversity may seem like a redundant exercise, but there is at least one prominent example of a AI being swayed by ugly ideologies.
In 2016, Microsoft launched an AI "chatter bot" via Twitter named Tay.
Tay was programmed to adapt its conversations as it learned from the humans tweeting at it.
But in less than 24 hours, online trolls had steered Tay into tweeting a series of highly offensive messages.
"bush did 9/11 and Hitler would have done a better job than the monkey we have now. donald trump is the only hope we've got," Tay said in one tweet.
Before Microsoft managed to delete the account, Tay had denied the Holocaust, advocated for genocide and called a games developer a "w---e".