The software, called “WatsonX AI,” includes governance tools that stop A.I. models from fabricating search results or providing users with offensive material, a phenomenon known in tech circles as “hallucinating.”
IBM said its new software and guardrails for A.I. will be made widely available starting on Dec. 5. The platform will help companies using A.I. platforms such as ChatGPT and others to ensure that their results are unbiased, factually correct, and explainable.
IBM noted that A.I. applications often rely on information and data from the internet that cannot be verified. The company said the new software is intended to help developers and consumers better manage the risks associated with A.I.
In October of this year, U.S. President Joe Biden introduced a sweeping executive order intended to boost government oversight of A.I., ushering in an opportunity for companies such as IBM to help resolve some of the output problems with A.I. models.
IBM’s stock has gained 5% this year and currently trades at $148.10 U.S. per share.