International Courant
An AI chatbot launched by the New York Metropolis authorities to assist enterprise house owners entry pertinent data has been spouting falsehoods, at instances even misinforming customers about actions which might be in opposition to the legislation, in line with a report from The Markup. The report, which was co-published with the native nonprofit newsrooms Documented and The Metropolis, consists of quite a few examples of inaccuracies within the chatbot’s responses to questions referring to housing insurance policies, employees’ rights and different subjects.
Mayor Adams’ administration launched the chatbot in October as an addition to the MyCity portal, which launched in March 2023 as “a one-stop store for metropolis companies and advantages.” The chatbot, powered by Microsoft’s Azure AI, is geared toward present and aspiring enterprise house owners, and was billed as a supply of “actionable and trusted data” that comes instantly from the town authorities’s websites. However it’s a pilot program, and a disclaimer on the web site notes that it “could sometimes produce incorrect, dangerous or biased content material.”
In The Markup’s exams, the chatbot repeatedly offered incorrect data. In response to the query, “Can I make my retailer cashless?”, for instance, it replied, “Sure, you may make your retailer cashless in New York Metropolis” — even supposing New York Metropolis banned cashless shops in 2020. The report exhibits the chatbot additionally responded incorrectly about whether or not employers can take their employees’ suggestions, whether or not landlords have to just accept part 8 vouchers or tenants on rental help, and whether or not companies have to tell employees of scheduling adjustments. A housing coverage knowledgeable who spoke to The Markup known as the chatbot “dangerously inaccurate” at its worst.
The town has indicated that the chatbot remains to be a piece in progress. In a press release to Engadget, Leslie Brown, a spokesperson for the NYC Workplace of Expertise and Innovation, stated, “According to the town’s key ideas of reliability and transparency round AI, the location informs customers the clearly marked pilot beta product ought to solely be used for business-related content material, tells customers there are potential dangers, and encourages them by way of disclaimer to each double-check its responses with the offered hyperlinks and never use them as an alternative choice to skilled recommendation.”
“The location has already offered hundreds of individuals with well timed, correct solutions and affords a suggestions possibility to assist drive steady enhancements of the beta software,” Brown stated. “We are going to proceed to deal with upgrading this software in order that we will higher help small companies throughout the town.”
Replace, March 31 2024, 9:23AM ET: This story has been up to date to incorporate a press release from the NYC Workplace of Expertise and Innovation.
NYC’s enterprise chatbot is reportedly doling out ‘dangerously inaccurate’ data
World Information,Subsequent Massive Factor in Public Knowledg
