Wednesday, March 29Welcome

Microsoft is adding OpenAI writing technology to Office

Microsoft 365

AppleInsider may earn affiliate commissions on purchases made through links on our site.

By incorporating OpenAI’s AI technology into its own apps, Microsoft may give Office users a way to create text for their projects.

AI has become more and more pervasive in creative fields, with tools like Stable Diffusion creating artwork from simple prompts. Microsoft is now looking into using the same technology to allow customers to insert characters into their pages and presentations.

According to sources with direct knowledge of Microsoft’s plans, informationMicrosoft wants to bring the artificial intelligence created by OpenAI into Word, Outlook, Powerpoint, and other apps.

For the user, this allows the document to be fleshed out with auto-generated stretches of text, all based on prompts. This includes her AI-generated emails crafted according to what the user wants to convey to the recipient.

It’s been years since Microsoft invested $1 billion in OpenAI in 2019 to work on new technology for its products. Little information has been published since then, but in 2022 Microsoft released Copilot, a coding assistance tool.

On Tuesday, it was reported that Microsoft could use OpenAI’s ChatGPT to provide natural language search results in Bing instead of lists of links.

Future plans may include integrating the same types of tools into the Office productivity suite, now known as Microsoft 365, according to current and former Microsoft employees.

Such efforts include improving Outlook email search results, such as determining which emails a user might be searching for without the user having to enter specific keywords. . It may also incorporate automatic replies to emails, recommendations for improving writing style, and other chatbot-style programs.

Privacy and Accuracy

However, Microsoft has to clear several hurdles, including correctness, for a successful implementation. Since ChatGPT was introduced, it has often produced results that I thought were believable and plausible, but partly because I haven’t been scraping the internet continuously for news and updates, it’s actually been in some form. I was wrong.

If Microsoft introduces AI text generation tools that provide false or offensive information, customers may stop using them in the future.

A further issue is that of privacy, as AI models must be safely customized to individual clients. It’s possible, sources claim, and we’ve already seen some success with the former in Microsoft products.

By preserving privacy and enabling client-specific learning of the model, the model recognizes individual or company-specific language, such as person or project names, and recognizes commonly used patterns of phrases and paragraphs. be able to learn. This information can be used to better understand the data ingested and improve the suggestions.

Microsoft is looking to add this technology to its products, but so far it’s done it in a variety of smaller ways, such as incorporating GPT into Word’s autocomplete feature. There’s no clear timeline for when the bigger push to make it dependent on will happen, but it could very well be within a few years.

Source link

Leave a Reply

Your email address will not be published. Required fields are marked *