‘Yes, I agree’ ‘Sounds good!’ ‘Great!’
以下内容由机器翻译生成。如果您觉得可读性不好, 请阅读原文或 点击这里.
Google is bringing its Smart Reply feature to Hangouts Chat for G Suite, its workplace messaging platform. The feature is similar to Smart Reply in Gmail, and uses machine learning to propose three different replies to conversations happening in Chat. The feature is rolling out now to workplaces that are either now running extremely efficiently or have halted to a standstill.
Google is determined to put its intelligent assistive writing tools to good use in its products. There’s Smart Compose in Gmail, which auto-suggests phrases within the body of your email as you’re writing, and Smart Reply in Gmail, which provides short phrases you can click to automatically respond to emails. These tools are going to start showing up everywhere, and it’s a sign that Google is pushing AI as an essential part of business.
Smart Reply in particular has been under some scrutiny for suggesting responses that are inappropriate or don’t quite grasp the tone and context of an email, and it’ll be interesting to see the tool put to use on a chat platform, where there’s even less characters to convey a message. The way we text says a lot about our personality, and everything from capitalization to adding a period to the end of your sentences is a choice that matters to some people. “Sure.” sounds a lot different than the more aloof “sure,” and it could mean the difference between someone thinking you’re mad, and someone thinking you’re very chill and cool.
Today’s news is only applicable to Hangouts Chat for G suite, so you probably won’t be seeing this in your Gmail anytime soon. However, Hangouts Chat for consumers has been in a state of transition since it was rumored that the service would be ending in 2020, as Google tries to figure out what to do with its many messaging apps. The company has clarified that Hangouts Chat will be available to consumers, but it might be a different product from Hangouts as it exists today.
We’ve reached out to Google for clarification on whether the feature will be pulling replies from the user’s own voice, or from the entire team’s language usage. Please humor me on this extreme thought experiment, but I imagine if the tool collects enough data on an entire workplace’s conversational habits, eventually it could be knowledgeable enough where people could just be clicking buttons at each other and create a platform entirely of bots just talking to bots. Again, that would either make for a supremely efficient business or an entirely incompetent one.