Amazon’s AWS appeals to corporate customers with new chatbot, AI safety measures By Reuters

© Reuters. FILE PHOTO: The brand of Amazon is seen on the firm’s logistics centre in Boves, France, October 6, 2021 REUTERS/Pascal Rossignol/File Picture

By Greg Bensinger

(Reuters) -Amazon is making an attempt to lure large company prospects to it AWS cloud computing service with a brand new chatbot for companies, and by providing to protect them towards authorized and reputational harm that may come from the output of synthetic intelligence.

The brand new chatbot, known as Q, is designed to assist with productiveness by serving to staff summarize vital paperwork and assist tickets and chat by way of communication apps akin to Slack, the corporate introduced at its annual cloud computing convention Tuesday in Las Vegas. The software program can even routinely make adjustments to companies supply code, rushing growth, the corporate mentioned.

The brand new software program arrives roughly a 12 months after OpenAI’s ChatGPT burst onto the scene, setting off a frenzy of funding in generative AI startups. Alphabet (NASDAQ:) and others have introduced their very own chatbots, which may have human-like conversations with customers to assist with every day duties.

AWS CEO Adam Selipsky, at Amazon’s annual cloud computing convention in Las Vegas, introduced a brand new safeguard towards objectionable content material on generative AI purposes, known as Guardrails for Bedrock. The service permits customers to filter out dangerous content material, he mentioned.

As a result of generative AI is educated on publicly obtainable content material, offensive phrases or different objectionable content material can slip by way of into outcomes from customers’ prompts. That’s notably problematic for youthful customers, in occasions of world battle or throughout elections when generative AI’s output in search outcomes can affect opinion.

Security advocates have cautioned that generative AI may function out of the management of its human creators and pump out more and more harmful content material or function whole techniques with out oversight. Specifically, they fear concerning the software program placing influential – and convincing – content material on social media websites like X and Fb (NASDAQ:).

Selipsky mentioned the brand new service was vital for patrons to place limits they see match on the generative AI they use.

“For instance, a financial institution may configure a web-based assistant to chorus from offering funding recommendation,” mentioned Selipsky. “Or, to forestall inappropriate content material, an e-commerce website may make sure that its on-line assistant doesn’t use hate speech or insults.”

As a part of its enchantment to firms, Amazon (NASDAQ:) mentioned the Q chatbot will supply companies limits in order that it could preserve delicate information from staff who mustn’t have entry to it. Pricing will begin at $20 per consumer, per 12 months.

Additionally on the convention, Amazon introduced it might indemnify its prospects towards lawsuits primarily based on the misuse of copyrighted supplies. Inventory pictures firm Getty Photos, as an example, sued Stability AI earlier this 12 months, alleging it scraped its web site for photos with out permission.

Guardrails for Bedrock is in restricted preview as we speak, Amazon mentioned. The Seattle firm didn’t present further particulars about its indemnification coverage.