Madison-based software startup DataChat has launched a new artificial intelligence tool called Ask.
This platform relies on large language model-based technology, a type of algorithm that’s been trained on massive datasets to understand and respond to human language. Users of this tool can type in questions about their own data and receive insights, the company says.
The announcement compares Ask to ChatGPT, a popular generative AI tool that has sparked conversations about the impacts of artificial intelligence on society since launching late last year.
Jignesh Patel, co-founder and CEO of DataChat, says large language model — or LLM — programs can’t always replace human intelligence and intuition “despite their impressive abilities.”
“With this in mind, we’re taking a human-centric approach to AI in data analytics by providing users with the option to provide feedback, rely solely on the output, or iterate on the analytics recipe with other human experts,” Patel (pictured here) said.
Along with that focus on the role of human interaction and control in data analytics, the company says ensuring transparency, reliability and data privacy are key challenges when applying AI in this space. The query responses provided by Ask include a “recipe” detailing steps the program took to reach its findings, which can be edited, replayed and saved, the release shows.
DataChat co-founder and Ask lead Rogers Jeffrey Leo John says that allows users to “easily iterate on their results,” replicate the analysis on different datasets and validate the AI’s approach.
“In many organizations, this type of self-documentation is critical to validate and audit decisions,” he said.
See more on the new tool here: https://docs.datachat.ai/work-with-your-data/ask
Listen to an earlier podcast with Patel: https://www.wisbusiness.com/2018/wisbusiness-the-podcast-with-jignesh-patel-uw-madison-professor-of-computer-sciences/
