According to news on December 29, in 2023, ChatGPT and other artificial intelligence chatbots can already generate realistic images and text that reads like human creation, shocking the world. AI experts predict that by 2024, artificial intelligence tools will be more powerful and regulations will increase, continuing to change the way we interact with technology.
Earlier this month, the NeurIPS conference was held in New Orleans, USA, with thousands of AI researchers and industry executives participating. They discussed where the industry is headed after a big year in 2023.
Aravind Srinivas, co-founder and CEO of the startup Perplexity, believes that in 2024 a company will release an artificial intelligence software whose performance will be comparable to or even better than GPT-4. GPT-4 is a large-scale language model developed by OpenAI and is the basic supporting technology of ChatGPT. Srinivas predicts that this model will come from one of OpenAI's many competitors.
In the next year, new artificial intelligence products will not only be more powerful, but also provide new application scenarios. For example, AI agents might perform more complex tasks with limited human intervention, such as making travel plans or ordering supplies. At the same time, other artificial intelligence products will transcend the limitations of text-based prompts and achieve more diverse inputs and outputs.
Oriol Vinyals, vice president of research at Google DeepMind, is the technical lead of Gemini, the company’s newly released large artificial intelligence model. He would like to see more AI systems that can react to various combinations of inputs, including text, speech and images, and generate different types of outputs. Google's demo at the Gemini launch event showed what's possible.
Margaret Mitchell, chief ethics scientist at startup HuggingFace, said that as technology continues to advance, movies and music performances entirely generated by artificial intelligence may appear next year. At the same time, the problem of autonomous bots inciting hatred and spreading misinformation online will intensify. This will prompt greater public scrutiny of the impacts and harms of artificial intelligence.
“These AI systems are definitely going to cause more disputes or lawsuits,” Mitchell said. Signs of that are already starting to emerge. The New York Times sued OpenAI and Microsoft on Wednesday, accusing them of using its content to train artificial intelligence services.
In addition, many celebrities, including Pulitzer Prize winner Michael Chabon and comedian Sarah Silverman, have previously sued OpenAI and Meta for similar reasons. Earlier this month, a British court ruled that Getty Images’ copyright lawsuit against StabilityAI could proceed to trial.
OpenAI has agreements with major publishers such as Axel Springer SE and The Associated Press. Such deals are likely to continue to increase in 2024.
Regulatory rules may also be introduced next year. Abeba Birhane, a cognitive scientist and senior advisor for AI accountability at the Mozilla Foundation, said she expects more AI legislation to emerge in 2024, but she worries it will be heavily influenced by corporate interests. In the United States in particular, lawmakers have met with AI executives multiple times to discuss policy. "Be careful of entangled interests when adopting their suggestions," she said.
While artificial intelligence will become more advanced and demonstrate disruption in some industries, Mitchell expects the intense hype of the past year to subside somewhat in 2024 as people realize that these tools are not always as magical as they initially seemed. “My guess is that the pendulum in artificial intelligence is going to swing,” she said. “People will stop being in awe of large generative models and treat them more rationally.”