Technology
Artificial Intelligence Work Assistants Struggle to Fulfill Promises
Jun
Artificial intelligence work assistants were crafted to offer businesses an accessible gateway to cutting-edge technology. However, this vision hasn’t quite materialized, with chief information officers noting it demands significant internal effort to extract full value from these costly tools.
Implementation Challenges
“It’s been more demanding than expected,” remarked Sharon Mandell, chief information officer at Juniper Networks, currently assessing tools from multiple vendors without immediate plans for deployment.
“It’s been more demanding than expected,” noted Sharon Mandell, Juniper Networks’ CIO, evaluating tools from various vendors, acording to Barron’s Print Edition
Promises vs. Reality
Products like Microsoft 365’s Copilot and Google Workspace’s Gemini aim to democratize AI assistant capabilities, offering secure, pre-packaged solutions for enterprises. Integrated with Microsoft and Google suites and extensive enterprise datasets, including emails and documents, they promise to reliably answer queries such as “what are our latest sales figures?”
Yet, reality often falls short. Data accessed isn’t always current or accurate, and the tools themselves are still evolving. Mandell observed that queries about 2024 data might yield responses based on 2023 data. Similarly, at Cargill, an AI tool couldn’t correctly identify members of the executive team, while Eli Lilly encountered inaccuracies in expense policy queries.
Learning Curve and Setbacks
These setbacks arise amidst growing corporate enthusiasm for generative AI, as CIOs explore potential productivity gains despite tools costing up to $30 per user per month. I remain optimistic about AI and believe we’ll achieve success. It’s just taking longer than expected, Mandell affirmed.
Data Cleanup Imperative
CIOs focus on cleaning and managing data to optimize AI capabilities. Bala Krishnapillai from Hitachi Americas highlighted challenges with inconsistent and duplicated data, which can lead to conflicting AI outputs. His organization continuously updates and refines data to maintain accurate “golden records” free of contradictions or duplicates. Kyndryl CIO Michael Bradshaw emphasized that data cleanup was their top priority, stressing the need for organized, current, and secure data to effectively leverage tools like Copilot.
New York Times sues Microsoft and OpenAI for copyright
In a surprising turn of events, The New York Times has taken legal action against tech giants Microsoft and OpenAI…
Vendor Responses
Vendors acknowledge these challenges. Microsoft’s Jared Spataro noted that as companies adopt Copilot, they discover overlooked or outdated data, prompting a realization of the need for more preparation.
To address these issues, Microsoft introduced Copilot Studio, enabling companies to guide queries to authoritative data sources within their systems. Spataro admitted challenges in prompting, emphasizing the importance of context clarity for optimal AI responses.
“Many users are encountering this technology for the first time and may feel somewhat disappointed,” Spataro acknowledged.
Richard Seroter, Google Cloud Chief Evangelist, highlighted the importance of Gemini for Google Workspace. This tool encourages better data management practices in organizations.
“If your data isn’t organized, AI won’t deliver its full potential,” Seroter cautioned.
Transforming Business Practices
Harnessing AI’s full potential demands more than technology purchases; it necessitates transforming business practices. This ensures data readiness and clarity in AI interactions.
Get a 3-year subscription to The WSJ All Digital with a 70% discount. No hidden fees or auto-renewals. Accessible on PC, Mac, smartphones, and tablets. Sign up now for premium content within 24 hours.
