Spend a lot of time using ChatGPT, or other AI chatbots, and it won’t take you long to realize that they spew out untruths.
Described as hallucinations, false information, or fabrications, these have now become a problem for every business, organization, and high school student who use artificial intelligence generative systems to draft documents. Some are using AI systems for tasks of potentially far-reaching consequences, ranging from psychotherapy to research to drafting legal documents.
"I don’t think you can find a model today who doesn’t suffer from hallucinations," said Daniela Amodei, co-founder and head of Anthropic, creator of the Claude 2 chatbot.
"[‘Chatbots’] are designed to predict the next word," Amodei said. "Therefore, there is a degree to which the model does this inaccurately."
Anthropic, the founding company of ChatGPT, OpenAI, and other major developers of artificial intelligence systems, also known as large language models, have said they are working to make these systems more reliable.
How long that will take – and whether they will be good enough to be considered safe for medical advice – remains to be seen.
"It can’t be fixed," said Emily Bender, professor of linguistics and head of the Computational Linguistics Laboratory at the University of Washington. "It’s a mismatch between the technology and the cases in which the technology is used."
Many rely on the reliability of artificial intelligence generation technology. The McKinsey World Institute predicts that this technology will add the equivalent of $2.6 trillion to $4.4 trillion to the world economy. "Chatbots" are just one part of this frenzy, which also includes technology that can generate new images, videos, music and computer code. Almost all of these tools include some language component.
Google has already started introducing an artificial intelligence product that writes news to media organizations, and for this, accuracy is paramount. The Associated Press is also considering using the technology as part of a partnership with OpenAI, which is paying to use some of the Associated Press’ text archive to improve its artificial intelligence system.In partnership with hotel management institutes in India, computer scientist Ganesh Bagler has been working for years to get artificial intelligence systems, […]
Artificial Intelligence sometimes invents things – can this problem be fixed?
