Community Forum – How are you currently using AI?

A: How are you currently using AI?

Please find below the responses to this IRI Community Forum Survey.

From an IRI member: We are looking to separate the reality from the hype to better understand how other companies are using AI. Thank you to those who took the time to share the tools you are using and what you’re using them for so we can all level set in this fast-changing space.  

Community Responses

Other:

• AI embedded in physics informed nueral nets, plus AI to predict tech trends (getFocus) as well as many different platforms including Palantir, Cognite.
• Grok
• Emerson global chatbot
• GrowthSignals; IPRally

Other:

• Everyday workflow augmentation (private assistant task types)
• Mostly emails. Would love to use agentic AI on our datasets.
• Searching internal reports for quicker understanding of what has been done before.

Other:

• Administrative productivity
• Not automation, but speeding up routine tasks
• Helps me craft efficient and clear communications to internal and external audiences with limited effort.

How do you separate what AI promises from what it can reliably deliver today?

• The missing link to date is the amount of upfront work (e.g. data structure and cleanup) required to take advantage of the promises made by AI.
• Currently sometimes a “trust but verify” kind of approach is needed with healthy skepticism that it may be needed for a while.
• It is a tool. Like all tools, it is not perfect for every task. The more people use it the better they are at understanding its capabilities.
• View AI as augmentation and junior contributors, so always have people review and signoff.
• AI promises much more than what it can deliver. In extensive searches it proves to be very effective (LLM).
• Very careful review of outputs.  Leverage it at an early stage when a framework is needed.  Iterate.  Review again prior to sharing output. 
• Setting expectations and communication/vetting is key.
• We emphasize the critical need for verification in all our use cases.
• Running it in real world applications to generate millions of simulations and show us scenarios never conceived before using slower models.