LinkedIn’s Smart Replies is a great example of how to get AI right

Linkedin AI based smart replies
Linkedin AI based smart replies

Well, these are the options that show up on LinkedIn’s messaging feature (a.k.a smart replies) and it does save a lot of time for me.

In fact, I don’t reply to 90% of LinkedIn messages – primarily because they are misdirected or too spammy; but because of LinkedIn’s smart replies features, I see myself atleast responding to ‘Hi Ashish’ sort of messages and in a way, carry the conversation.

What’s right about this implementation of AI in Linked’s Smart Replies feature?

  1. Unlike other bots, it doesn’t obstruct the flow. If you wanna use it, use it. Else, move on. That is, the ‘smartness’ is limited to your context which you can override.
  2. It is neither over promising nor under delivering. Most of the AI based chatbots/features tend to be a ‘solution for all your queries’ and eventually they fail very badly (look at Facebook M).
  3. LinkedIn isn’t trying to prove how deeply they understand AI. This isn’t a case study (unlike Facebook) to drive AI narrative. They have just launched a smart feature and it does seem to work. This is great example of product management knowing what (not) to do.

“We had a problem to try and solve: How can we give time back to busy people? There were two main things we thought about with this. First, we wanted to keep the phrases short and simple. Things like ‘Sounds good,’ or What time?’ are simple enough that it’d be impossible to say if a human or a machine wrote them. Second, we wanted the UI to be unobtrusive. If the user wants to use ignore Smart Replies, or build a longer message, using Smart Replies as a jumping off point, they can do that.” [Arpit Dhariwal, a Senior Product Manager at LinkedIn/source]

What’s your take?

Why Product Management Should not be Responsible for Project Management

Android Fragmentation: Oreo is at 0.7%; Kitkat (still) at 12.8%