And now some of the negative ones.
McDonald’s is removing that AI-powered Automated Order Taker technology from over 100 drive-thrus after videos showing flaws went viral. The technology was developed in collaboration with IBM, but customers reported issues with incorrect orders. McDonald’s will continue to explore technology solutions for its workflow and evaluate a future voice-ordering solution by the end of the year.
A study by researchers at Stanford University finds that AI legal research tools, despite claims by providers, still suffer from a significant rate of hallucinations or demonstrably false outputs. The study tested products from major legal research providers and compared them to OpenAI’s GPT-4 on over 200 legal queries. It found that while hallucinations were reduced compared to general-purpose chatbots, the legal AI tools still hallucinated at an alarmingly high rate. The researchers emphasize the need for transparency and benchmarking in legal AI to mitigate these issues.
According to research published by Pegasystems, business leaders need a clearer understanding of generative AI, leading to failed implementation projects. Despite claiming to have a good understanding of AI, leaders overestimate their knowledge and underestimate the time AI has been used. Decision-makers are willing to invest a significant portion of their IT budget in AI solutions but acknowledge wastage due to a lack of a well-defined AI strategy.
SolarWinds’ 2024 IT Trends Report on artificial intelligence (AI) reveals that while IT professionals strongly desire to adopt AI technology, they have concerns about data quality, infrastructure readiness, and security. The report emphasizes the need for internal systems, prioritizing security, fairness, and transparency in AI development. IT professionals see AIOps as the AI technology with the most positive impact but distrust the data powering AI and cite privacy and security concerns as significant barriers to adoption. They also call for government regulation and the development of thorough ethics, data privacy, and compliance policies for successful and secure AI adoption.
Replacing humans is a bad idea. The failure of McDonald’s AI-powered order taker underscores the importance of reliability and accuracy in customer-facing applications. IT service providers must prioritize thorough testing and continuous improvement to avoid similar pitfalls. That doesn’t mean it’s not a good idea to try.
IT service providers need to address foundational issues such as data quality and infrastructure readiness to facilitate successful AI adoption. Ensuring robust, secure, and scalable infrastructure is crucial for leveraging AI effectively. It remains the key opportunity.