Want Trustworthy AI? Can You Trust Your Data?
Interesting tidbit from this study by Protivity:
Among organizations in which ROI on AI investments has significantly exceeded expectations, nearly all (97%) are confident in their ability to obtain, organize and understand the data required to support their AI goals.
We’ve seen a lot of discussion about AI projects not showing significant ROI, and this may be because the results are not trustworthy, which raises the question of whether the data being modeled is reliable. If AI is surfacing incorrect information because the model was built on outdated and inaccurate data, the problem lies with the source of the data.
This study shows us how a successful AI rollout in business isn’t about finding the magic tool that will make sense of all of your data, it’s also about making sure your data is clean. That’s a very different project rollout, one that won’t come from your IT Department. It’s not just a technology issue.
As I type this, I am fully aware that the internet is full of crap data. Scraping the ever-increasing amount of fake news into an LLM isn’t going to make it more trustworthy. I’m not sure that public AI can ever be trusted, which makes me question how consumer AI products will ever return the investment that has been made.
Follow these topics: Artificial Intelligence
