Artificial intelligence could create more work for staff, study finds

Artificial intelligence could be the reply, in accordance with a brand new study by the Australian Securities and Investments Commission (ASIC) and Amazon. create more work for staff after researchers used Meta’s open-source Llama2-70B software to perform business tasks within the workplace.

According to artificial intelligence (AI) programs and a bunch of 10 individuals were asked to summarize submissions specializing in ASICs, recommendations, and regulatory references. In addition, the tool was asked to incorporate each context and page references. Both sets of responses were then blindly scored by reviewers for consistency, length, and identification of ASIC references, in addition to regulatory references.

The AI ​​programs fared poorly in comparison with their human counterparts, scoring 47% in comparison with humans’ 81%. According to the team that conducted the study, the AI ​​performed particularly poorly when it got here to finding references to AISC within the documents it was alleged to use to create its summaries.

“Finding references in larger documents is notoriously difficult for LLM due to context window limitations and embedding strategies,” the team wrote. “Page references are not traditionally stored in embedding models because the content of PDF documents is retrieved as plain text. To achieve greater accuracy in this problem, significant progress has been made by splitting documents into pages and treating pages as fragments with associated metadata.”

According to the study, reviewers often needed to “refer to the source material to confirm details of the AI ​​summary” and “generally agreed that the AI ​​results could potentially generate more work if they were used (in their current state), either because of the need to fact-check the results or because the source material better presented the information.”

While this study has limited application to AI, it does suggest that the expected advantages of AI could also be limited by the necessity for humans to do more work to satisfy the precise needs of AI within the workplace.

Several reports indicate that black people and artificial intelligence are at odds with one another, a bunch of researchers from Stanford University say provided some feedback to the Congressional Black Caucus within the March 2024 White Paper

According to the researchers, “While AI has the potential to exacerbate racial inequality, it can also benefit Black communities. If implemented carefully, AI has the power to improve access to healthcare and education, as well as create new economic opportunities. For example, AI can help doctors make more accurate diagnoses and provide personalized treatment plans, especially in underserved communities where access to healthcare is limited.”