Microsoft to Add GPT-3 and DALL-E AI Writing Assistant to Word, Powerpoint, Outlook, and Teams
New features to make your Office 365 and Teams subscriptions more valuable...and maybe increase you monthly bill
The Information has another scoop on Microsofts’s intention to employ OpenAI’s GPT-3 (or GPT-4 when it is ready) and DALL-E in its market-leading office productivity software. Granted, this is an idea that was not hard to anticipate. More business users write, create presentations, send emails, and conduct meetings with Microsoft 365 than any other suite of productivity tools. According to The Information:
Microsoft has discussed incorporating OpenAI’s artificial intelligence in Word, PowerPoint, Outlook and other apps so customers can automatically generate text using simple prompts, according to a person with direct knowledge of the effort.
This news dropped shortly after other rumors surfaced that Microsoft was likely to add conversational search features using OpenAI technology to the Bing search engine in early 2023. Microsoft’s $1 billion investment into OpenAI in 2019 included a collaboration on Azure hosting services and other unnamed services. Apparently, we will soon see the fruits of that collaboration in Microsoft applications.
Text-to-X Generative AI as a Feature
While there are many stand-alone text-to-x generate AI solutions on the market today, Synthedia has noted in the past that the fastest growing use of these technologies is as a feature for existing products with large user bases. Canva and Picsart added AI image generation features this past fall, and the latter claimed its users were creating a million images per day in November.
The biggest adopters of large language model (LLM) applications, such as AI writing assistants, have been pure-play solutions. It was only a matter of time before these features began to show up in the leading word-processing and other business software applications.
How Microsoft intends to use these capabilities is not entirely clear. Companies like Jasper AI have had great success offering full AI-enabled writing suites with numerous templates that generate entire documents for business professionals. On the other end of the spectrum, there are tools like Grammarly and AI21’s Wordtune that don’t write for the user but do make suggestions on how to improve the writing.
The latter seems like a more likely feature approach for Microsoft, but the company is considering a wide range of options. The Information reports:
“So far, Microsoft has worked on incorporating existing and future versions of OpenAI’s language-understanding model, GPT, to provide more useful search results when Outlook email customers look for information in their inboxes. For instance, GPT has the ability to figure out what emails the customer might be searching for even if they don’t type the exact keywords that are in the relevant emails, two people with direct knowledge of the plan said.
“Microsoft executives and researchers also recently looked at how Outlook and Word could use GPT to suggest automatic replies to other emails, or to recommend changes that remove jargon and make people’s writing more easily understandable to others, one person said. They also talked about using OpenAI’s software to create chatbot-style programs inside Word and Outlook that write entire tracts of text in response to a prompt, similar to OpenAI’s headline-grabbing ChatGPT experimental online chatbot, according to someone with direct knowledge of the discussions…
“It’s not clear whether Microsoft has established a specific product road map for incorporating OpenAI’s latest GPT models into Word, or when such features could become available. But the Word team has been eyeing how OpenAI’s models could beef up the app’s ability to understand and summarize documents, correct grammatical errors and suggest clearer syntax, the person said.”
What About Teams?
Microsoft Teams is also a strategic application for the company that can benefit from LLM applications. Online meetings may have transcripts that need to be created from spoken content. OpenAI’s Whisper speech recognition solution is showing promise for high accuracy, reasonable pricing compared to the leading cloud solutions, and support in multiple languages.
The meetings also may benefit from summarized notes and action items or the real-time surfacing of useful contextual information. Microsoft’s potential benefits extend beyond the Office suite and search.
Protecting the Brand and the Revenue
I just wrote about Google and its Innovator’s Dilemma around conversational search. Clayton Christensen introduced The Innovator's Dilemma in 1997 to explain why the incumbent market leaders are so often displaced by new entrants when disruptive technologies shift customer behaviors and preferences. The mistake of incumbents is that they typically try to protect their existing business models and assets for too long and wind up creating a path for a competitor to gain customers.
Microsoft and Google have an innovator’s dilemma problem from a business model standpoint. If things change too rapidly, their market-leading solutions may look out of step with market requirements. That could be very costly to these companies’ businesses that rake in enormous sums of cash.
The companies also need to consider protecting their reputation. They have vast businesses, and a public relations firestorm in one product line could undermine confidence across the portfolio. That is a risk most executives will avoid. You should expect a conservative approach because you never quite know what a generative AI model will create.
A paragraph with obscene or inappropriate content created by a Microsoft Word text-to-writing feature would generate negative market sentiment. Big companies like Microsoft are expected to have their act together. The same misstep by a startup will not receive as much media coverage because so few people know who they are. And startup companies are likely to be forgiven more quickly if they apologize and commit to doing better in the future. Microsoft would almost certainly be expected to conduct an analysis about how it happened, share the results publicly, and explain how they are going to prevent it from happening again. The standards are different, and therefore the risk management and risk tolerance are different.
This makes me think a writing assistant that offers suggestions as opposed to generating new written content from scratch is the more likely path for Microsoft at the outset. This also means the company doesn’t have to worry about GPT-3’s penchant for serving up incorrect and unverifiable information. If the AI writing assistant only helps business professionals augment and refine their own writing and research, then many of the risks with substantial downsides never materialize.
The New Must-Have Feature
What these moves also suggest is that so many solutions are going to have features for text and image generation they will quickly be considered requirements by users. The absence of these features may cause customer churn to other solutions.
Microsoft also has another reason to care deeply about generative AI beyond customer retention. The features are likely, at some point, to carry an added monthly subscription fee for their availability. Paying for access to these solutions is expected by professional users. That could drive higher revenue for Microsoft from their 365 and Teams products.