Since the launch of ChatGPT, AI has made mainstream media headlines week after week.
The momentum seems unstoppable, following another swirl of stories in the past seven days.
On that note, here are some of the most eye-catching AI scoops from the past week, each touching the customer experience space in one way or another.
Air AI Introduces a Bot That Handles 40-Minute-Long Customer Calls
In recent days, the following demo from little-known tech provider Air AI went viral.
The video shows a bot making a proactive, outbound support call, sounding human-like and engaging in small talk.
Sources suggest that the bot has infinite memory, perfect recall, and sparks actions across over 5,000 apps.
As an excited LinkedIn user wrote:
It’s basically like having 100,000 sales and customer service reps at the tap of a button 🤯 This will revolutionize entire industries.
Indeed, the demo is impressive. Yet – to somewhat dampen the fire – it’s critical to note that it is likely a demo, not a live call recording.
After all, they do not discuss the day of the test drive; they skip straight to the time.
Moreover, the bot never demonstrates an understanding of what the customer said. It acknowledges them and moves on to the next query.
As such, older techniques – such as entity extraction and paraphrasing – are seemingly at play.
Nonetheless, it’s a fun demo that underlines the strides businesses are making in conversational AI and natural voice in particular.
OpenAI Releases Code Interpreter
Code Interpreter is a ChatGPT plug-in that allows users to upload their own data. It then applies a built-in Python interpreter to run and analyze functions on that data.
From there, it’ll produce files for downloading based on those calculations - all within a sandboxed, firewalled container.
A set of Python packages also comes with Code Interpreter, allowing users to build charts, graphs, and other visualizations of their data. They may even make predictions.
Finally, there is a “show work” button, so the business may see the code it’s running.
All this may have significant implications for data & analytics professionals, who may begin to automate more complex workflows with ChatGPT.
Anthropic Unveils Claude 2
Claude is a GenAI bot built by Anthropic, following the premise of “constitutional AI”.
As such, AI trains its LLM by following a set of rules instead of relying on human feedback.
In doing so, it has produced some impressive results. Indeed, Shobhit Varshney, VP & Sr. Partner at IBM, says:
On real-world client use cases, [we] did some A/B testing between GPT and Claude 2 - super impressed. 4-5x cheaper than GPT4-32k, knowledge cutoff early 2023.
Alongside this, Claude 2 can handle 75,000 words as input, which is more than three times as much as GPT-4.
Moreover, users can upload multiple documents simultaneously and have Claude summarize them. Users may then ask questions about the content, which Claude can answer.
Interestingly, Zoom will soon embed Claude into its CCaaS platform – helping customers solve queries autonomously and agents find answers across systems.
Yet, this is just the start. Its new word-crunching capabilities may bring further workforce optimization (WFO) use cases to life.
IBM Releases WatsonX, A GenAI Platform
IBM launched WatsonX, an AI, data, and governance platform that allows enterprises to create “open, trusted, and targeted” generative AI.
In doing so, the enterprise tech giant strives to fill a "meaningful gap" in a competitive landscape.
