‘Gamechanger’: OpenAI launches ChatGPT4

OpenAI has released a powerful new version of artificial intelligence program ChatGPT, capable of passing the US bar exam and coding entire video games within minutes. GPT-4, launched March 17, is similar to its predecessor in that it uses generative artificial intelligence to use algorithms and predictive text to create content based on written prompts. In a technical report, Microsoft-backed OpenAI said the model had “more advanced reasoning skills” and could produce more nuanced and factual content than its predecessor. The San-Francisco based company also said the model could make “simple reasoning errors” or be “overly gullible in accepting obvious false statements from a user." Within hours of its release, users reported being able to create computer games in less than 60 seconds, generate websites from drawings, and build tools to help people identify and order medications. UNSW Artificial Intelligence Institute chief scientist Professor Toby Walsh said the developments were “just the beginning." “These tools will get better,” Professor Walsh told Campus Review. “This is a new way for us to interact with computing devices. “We can have this interface that we can have intelligent conversations with and it will remember things in the context of what we are saying and help to fill in the gaps.” As part of OpenAI’s GPT4 release, the company announced commercial partnerships with online learning companies Duolingo and Khan Academy to roll out personalised ChatGPT powered ‘AI tutors’. Professor Walsh said this is a significant example of how AI systems will revolutionise education in the future. “That's going to be a real gamechanger,” he said. “If you’re determined to get an A grade, you can ask ChatGPT if it’s good enough and get it to critique your essay and point out how to improve your grade." A recent survey commissioned by the University of Queensland found that only around 40 per cent of Australians trusted the use of AI in their workplaces. Over 70 per cent reported having concerns about cybersecurity, privacy, manipulation and misuse when interacting with AI-powered technology. Australia currently has no laws regulating the use of AI, which is expected to be worth $22.5 trillion to the global economy by 2030. Last week, federal industry and science minister Ed Husic announced the creation of the Responsible Artificial Intelligence Network - an 'world-first' body tasked to guide ethical and safe use of AI. The network was launched by the National AI Centre and will be coordinated by the CSIRO. Professor Walsh, who is a member, said Australia was "behind the rest of the pack" in AI regulation. "The Australian government doesn't seem to be very keen to regulate the tech space, full stop," he said. “This network is an industry body, it’s bringing together all of the major players in the industry to provide a community of practice. “But we’ve yet to see, for example, whether the national AI centre is going to invest resources in that.”