Autonomous You: the age of personal bot ambassadors (bottr.me review)

Julian Harris
Aug 28, 2017 · 4 min read

Hey everyone you can now train a bot to represent you in conversations so you don’t have to participate and get on with your life!

I thought I’d try it anyway. The TL;DR version is:

  1. Connect to social media: you connect it to all your social media accounts
  2. Analyse social media and build up a profile: It processes them and finds topics you seem to know something about, and rephrases them in the form of questions you can encourage users to ask. E.g. for me it suggested “what do you think of kingston council” (see below)
  3. Launch bot: it creates a url at bottr.me E.g. see mine.
  4. Promote: you then encourage people to interact with it with the assumption that it will be innocuous, professional, on-message for your personal brand and engage with you once the bot has decided the conversation is important enough to engage with the real you. I guess?

Looking more closely

Here’s the admin screen. You can see I’ve connected LinkedIn, Twitter, Google+ (well I am a Xoogler), and Medium. My use case is “I want to present a new way for people to get to know my professional self more quickly accurately and easily”. So yeah, no way I’m connecting Facebook as that’s purely 100% personal, and even Twitter is marginal as it has random stuff in there that probably isn’t helpful for my “professional brand”. But let’s see.

What I see when editing my bottr virtual me
What you see when you try and engage with my virtual me.

How did it fare? Too early to tell.

What do I think about Kingston council? Win: it picked up a tweet to our then-local MP on a great initiative around council expenditure. Nice!

Do I like music? Fail: I have listed in my Twitter profile that I’m an amateur musician, so that’s pretty surprising it’s not picked this up.

When was I born? That hasn’t been answered; it’s almost reassuring that it doesn’t know.

Where humans are involved is critical

I think “Autonomous You” is an inevitable attraction and as natural evolution from online profiles will see huge growth.

Firstly, the v1 of Autonomous You will continue to be around questions and answers. They’re simple, have limited context and still have potential to have improved usability over static digital profile you search against. In future as the quest for more engagement and safe ease of use increases we’ll see companies better at integrating dialog and bringing it into the professional dialog. Maybe the LinkedIn killer could be something like bottr.me but for pros. Update Nov 2017 Soul Machines with their life-like virtual assistants envision personal assistants within 5 years (currently creating avatars take 8 weeks and a team of specialists).

Controlling engagement with human throttles

Fully autonomous is perilous for the foreseeable future, as Tay and Zo have shown. The real answer for most will be ongoing operational processes for questions it can take and answers it can offer, along with thresholds for acceptable variations. For those who are more adventurous, reactive monitoring could show in an easy-to-use flow any variations that have been coming up, while those wanting to control their personal brand more will have people curate each and every variation on both sides.

What I expect to see is growing best practice around human throttles on the learning loop to include at least the items below:

  1. Question selection: How questions are discovered and curation (autonomous vs fully preapproved). Here I think bottr.me is controlling every question it can answer.
  2. Question variation selection: How flexible question variations can be (and how much control here too: preassessed by the bot as “very similar” to perhaps “ok” match)
  3. Answer selection: which answers to give to which questions. Again the lower-risk situations will give more autonomy, while higher impact situations will require a lot more control over what answers can be given.

Expect growth in ethics, compliance and abuse management

Also I fully expect a niche industry to grow to support ethical and abuse management objectives. Where chat bots involve free text I’d expect to see plugins that deal with important out-of-band critical use cases e.g. indications of self harm, and abusive or spamming behaviour. These specialists will be language independent and have a strong real time network of support resources to keep up to date with spam etc.

)

Julian Harris

Written by

Technical Product guy specialising in the impact of machine learning on speech and text (NLP). Ex-Google, but none of this matters if we don't fix the planet.

Welcome to a place where words matter. On Medium, smart voices and original ideas take center stage - with no ads in sight. Watch
Follow all the topics you care about, and we’ll deliver the best stories for you to your homepage and inbox. Explore
Get unlimited access to the best stories on Medium — and support writers while you’re at it. Just $5/month. Upgrade