AT&T hopped on the generative AI hype train, detailing plans to deploy OpenAI’s ChatGPT service among staff in a bid for improved efficiency and creativity.
The US operator plans to deploy Ask AT&T in the coming weeks, citing uses in software coding and development, translating customer and staff documents from English to other languages, and simplifying its records.
AT&T addressed the elephants in the room in terms of the AI’s security and accuracy.
It worked through potential security concerns with OpenAI backer Microsoft to make the set-up “safe for our employees and our corporate data”.
Ask AT&T runs in a “dedicated Azure tenant that’s been pressure-tested for leakage”, meaning staff can input company data and information “without worrying” about it becoming public.
Precision is ensured by employing in-house nous and processes “that focus the system on responsibly delivering accurate results”.
It noted generative AI tools “are great and getting better quickly, but they’re not magic or infallible”. Thus, the operator is keeping an eye on “external misuse” of the technology, developing countermeasures covering the set-up itself along with AI-generated attacks.
AT&T highlighted a long history of employing AI in its operations, a provenance which includes a claim to being present at a meeting in 1955 where the term artificial intelligence was coined.
More recently, AT&T has employed AI for customer service, operational efficiency and preventing spam.
Potential developments being assessed for Ask AT&T include network optimisation; updating legacy software; further customer service uses; HR queries; and cutting meeting times by automating the creation of summaries and action points.