metamorworks - stock.adobe.com
Microsoft, pursuing its strategy of integrating OpenAI's generative AI technology within enterprise application platforms, on Monday introduced Power Virtual Agents conversation booster and AI Builder model with content generation.
Power Virtual Agents enables developers to create AI-powered chatbots for different scenarios, such as a bot for answering complex questions or one to engage with customers in multiple languages. AI Builder is a Microsoft Power Platform feature that provides AI models that help automate business processes.
Power Virtual Agents conversation booster equips enterprises' chatbots with capabilities from GPT -- the large language model from the tech giant's partner, OpenAI -- to answer questions when connected to company-specific resources such as a public website and internal knowledge base.
In addition, AI Builder now includes Azure OpenAI services in its interface, giving users access to new low-code generative AI models and templates in Power Automate and Power Apps.
Generative AI capabilities
Given Microsoft's partnership with and $10 billion investment in OpenAI and its powerful large language model and chatbot, ChatGPT, bringing generative AI capabilities to Power Platform apps and AI Builder is a natural step, said Kashyap Kompella, CEO of RPA2AI Research.
"ChatGPT has massively raised the bar on expectations for enterprise bots," Kompella said.
When GPT-based bots are trained on a company's documents, it is easier to incorporate new information into customer support processes, he added.
Kashyap KompellaCEO, RPA2AI Research
Meanwhile, GPT-based text generation helps workers become more productive. Text generation can assist employees with tasks that usually take time, such as summarizing, rewriting and reformatting.
"These are early days yet, but by combining the conversational interfaces with foundational [large language model] LLM AI applications, every employee can have their own AI assistant to make them more productive on the job," Kompella said.
Conversation boosters enable enterprises to use their most current and valuable data sources to construct responses that their chatbots can give to a question it has yet to be trained on.
"What they're saying now is you won't have to predefine these conversations," Forrester analyst Mike Gualtieri said. "ChatGPT is smart enough to deal with many more situations without defining them."
However, while these features are geared toward making certain routine tasks easier for enterprises, Microsoft cautioned that the tools are experimental and their accuracy and relevance "cannot be guaranteed."
The experimental stance comes as LLMs such as GPT and Google Bard have proven unreliable in more than a few instances, with LLMs acting expressing anger, composing irrational ideas or just being notably inaccurate.
"We are not at the 'make business case and roll out at scale' phase yet," Kompella said. He added that while many companies are curious about experimenting with ChatGPT, enterprises should use caution.
"It's easy to break user trust with LLM applications, as they can be unpredictable and unreliable," Kompella said. "It's imperative to fully adhere to principles of responsible AI development."
Enterprises may be better off using the AI features as innovation projects to understand how they work within their business processes and employee work routines and notice any uptick in productivity, engagement and satisfaction measures, Kompella added.
Organizations may also consider being selective with the applications for which they employ these features, Gualtieri said. And internal applications will be safer than external ones, he noted.
Moreover, since the features are included in the Power Apps, developers will use them, giving enterprises an extra level of safety.
"I'm not saying it's perfect, but there's already a little bit of governance in this process," Gualtieri added.
Also, enterprises may want to disclose to consumers that their bots are based on ChatGPT.
"That's the great thing about ChatGPT and the widespread knowledge about it is that people kind of know what to expect, and they expect mistakes," Gualtieri continued.
The new GPT features are available in only the United States and in English. Enterprises can't publish their bots while the features are enabled, but they can be disabled in the settings anytime.