Send me Google's new 'Smart Reply' in a message and I will unfriend you

This article may contain personal views and opinion from the author.

Google unveiled a bunch of new things at I/O 2016, but the thing the new chief executive Sundar Pichai focused on was deep learning and artificial intelligence.

Being a leader in terms of the amount of data it gets from daily user queries, Google has the world's knowledge at its disposal and the company wants to do something more with it than just show you ads. We've already gotten a first look at the amazing powers of the Google knowledge graph when it launched Google Photos last year, an impressive service with magical search powers that is able to look through your images and - if you ask it - show only those with hugs, or only those of your dog, and so on. The ability to magically recognize what's on a photo has truly made it stand out.

At I/O 2016, Google is going one step further. It unveiled a new messenger called Allo that will use the same super-power: once you send an image to someone in a chat, the other person will get a bunch of 'smart replies' that Google will automatically suggest as it recognizes what's in the image.

But this goes a step too far.

I understand the convenience and the nerdy joy of having a computer show you responses to images and text rather than you having to text. Laziness is the father of innovation, as they say. And it could be genuinely useful: say you receive an image of tulips or some other, more exotic flower that you cannot recognize. With Smart Replies, you might get the name of the flower, which is nice.

However, in the promo video for Allo, Google showed someone sending a picture of a kid and the person on the other end of the chat got automated response options along the lines of "aww how cute".

What's wrong with that, you may ask, it's an appropriate response after all. Yes, but when I send something as dear to me as a picture of my kid, I do not want to receive an automatically generated response by my friend. Using 'Smart Replies' in that context is not only the opposite to smart - it's offensive to the close relationship that a friendship is supposed to be.

In that imaginable scenario, it's impossible to know whether the response that the other sends is automated, but if my friend sends me a 'Smart Reply' in that situation, I will unfriend them.

Programmers keep on forgetting that the human condition is a lot about... inefficiency. The few seconds or minutes that my friends spends thinking about the right way to express their feelings about that dear picture I just sent are not the most efficient way to spend your time, but contrary to logic, that's what we, humans, value.

I can't find the exact slide, but I remember clearly another example given to illustrate the power of technology. Someone had set a reminder to 'call mom' on their smartwatch. Quite honestly, if you can't remember to call your mother and need a reminder for that, you have most likely failed at being a human. Sure, there are all sorts of families and all sorts of family relations, but yet still.

At the end of the day, technology is about enriching our lives, making them better. Sometimes, we have to stop and ask ourselves: is that latest innovation doing that? In the case of 'Smart Replies' that have the potential to turn our human conversations in a bot-generated scripts, I'm afraid that the answer is not necessarily a 'yes'.



1. SuperAndroidEvo

Posts: 4888; Member since: Apr 15, 2011

But people all the time text just like this. No one ever uses complete sentences or even proper punctuation. So how would you know if it's bot generated or just some common response? Victor, how will you know when everyone texts like that? I believe I am the only person in this world who actually writes a complete sentence with proper punctuation when I send a text message. It must have been that my Mother was an English teacher when I was growing up.

3. Victor.H

Posts: 1099; Member since: May 27, 2011

Do they? I don't know. There is no way to know whether what you get is an automated response, but imagine how little the person on the other end of the line cares if he won't even bother typing a proper response. The point is not about the response being with proper grammar, I prefer an honest and emotional 'Whaaaaat' or 'WTF!!!!' or something along those lines in some cases rather than an indifferent bot-generated response.

8. SuperAndroidEvo

Posts: 4888; Member since: Apr 15, 2011

What is scary is that Google's "Bot" generated responses are actually 100% common. It's pretty much how people text. So if there is something that you would normally type out, & it's already there then just use it. Me I wouldn't ever use them because like I stated before, I actually take the time to swipe out a complete sentence with a proper punctuation. It's not that I am being pretentious, it's just the product of having an English teacher be your Mother. Now being a parent I don't ever want my kids to type like that. I want them to use proper sentences because the English language or any language as a matter of fact is being eroded. My wife is an underwriter and she sees agents and other employees actually using a "text" short hand form of writing on actual documents that could be used in a court of law. That makes the company look foolish. Now the corporation she works for needed to send a memo to all employees so they use proper writing instead of common short hand "text" style of writing. It's crazy but it is happening out there. I don't want my kids to grow up thinking it's OK to always write in the style of short hand text. Could you imagine a student writing a book report in the short hand text style?

9. tango_charlie

Posts: 347; Member since: Nov 16, 2011

But you seem to forget one point: The actual response is not really just bot-generated. It makes a suggestion which you can choose from. The same does your brain when you have to reply, which costs you cognitive resources to think about the right response options. Now you get this response options without the need to think about them and you don't have to use them IF they don't suit your thoughts/emotions. it's just a cognitive support for "modern" sensory overloaded mankind.

18. talon95

Posts: 1004; Member since: Jul 31, 2012

It's not "indifferent bot generated". It learns what you like to say and provides your own vocabulary. If seeing how poor someone's vocabulary is shocks them into writing a full sentence response then you should like it either way. I'm not all that into the new bot era, but I wouldn't fault it yet.

4. 444777

Posts: 399; Member since: May 02, 2016

Where have you been....? It's been long time you haven't posted? Miss you're posts.

12. SuperAndroidEvo

Posts: 4888; Member since: Apr 15, 2011

I always come to It's just that I don't post as much as I really want to. Victor though got me to post, because he thinks the automated responses are no good. Now if this is really AI based and I always use sentences with proper punctuation will my choices be something I would write? That is my biggest question with Allo, because if it really delivers then I would find this feature rather useful.

21. marorun

Posts: 5029; Member since: Mar 30, 2015

If i understood its will start with generic one but learn about you over time so yeah in time its will give what you would say.

26. SuperAndroidEvo

Posts: 4888; Member since: Apr 15, 2011

Well I won't use the generic stuff, hopefully the AI picks up quickly with me. +1

17. femiRicHkiDD

Posts: 15; Member since: Aug 21, 2015

And the Rude part was he used an iPhone sketch for a Google relates news ? Maybe he doesn't have an idea how a nexus phone looks like *smh*

30. SmartNingen

Posts: 48; Member since: Apr 02, 2015

I fail to understand how that is rude. Google Allow will be available on both iOS and Android.

31. SmartNingen

Posts: 48; Member since: Apr 02, 2015

"I believe I am the only person in this world who actually writes a complete sentence with proper punctuation when I send a text message." Damn. I used to think that too. Now that I've met you though, I know I am not alone. My mum isn't an English Teacher though, I just love being a Grammar Nazi.

35. andynaija

Posts: 1265; Member since: Sep 08, 2012

Same here, I'm just used to writing/speaking proper English.

37. JC557

Posts: 1925; Member since: Dec 07, 2011

I would type out a complete message now that we're no longer using the dial pad to create a text message.

2. Tomgabriele

Posts: 30; Member since: Feb 25, 2016

Relax! There is not some magical threshold of what should and shouldn't be automated. The user is still choosing how to respond; if their communication is stunted because they only use suggested replies, it's their fault, not the tool's fault. Besides, the AI learns how the user talks, so generated responses won't be identifiable anyway.

5. jroc74

Posts: 6023; Member since: Dec 30, 2010

And I thought you was an Android fanboy.... As the years go by its all about efficiency. Everywhere I look someone has a smartphone, and the users are getting younger n younger. But to your point.....what about the ppl that text more than use the phone? I had some co workers like that. Real story: I had a co worker who had a relative pass away. I found out by seeing posts on Facebook. I decided I would let them have their grieving moment and pay my respects when I saw them in person at work. That didnt go over too well....they were highly upset...because I didnt post on FB or call but waited until I saw them in person. It was literally the next day. When I choose a phone its for battery life and reception first, everything else 2nd. So I talk on the phone more than text and I barely use FB. This same co worker will rarely answer their phone but will respond fast when you text. I understand where you are coming from.

6. realjjj

Posts: 375; Member since: Jan 28, 2014

What are you 12? It suggests a reply to save you from typing and if it's good enough it will know what you would say. Maybe learn to think insted of being a whiny child and copying some idea you've seen from some other whiny child elsewhere.

7. tango_charlie

Posts: 347; Member since: Nov 16, 2011

Using test message is about efficiency. Replies have to be short and concisely. Humans are worse in multitasking and in most cases when we replying to someone/thing we are doing something else: being at work, school/university; watching tv, doing homework, shopping, talking with someone face-to-face...So we have neither the time nor in that situation the cognitive resources to think about an adequate response, so computer generated responses come quite handy. It's not like you get an automatic response, cause you still have to choose between the responses, so it's still your personal choice. It's like being hungry and don't know what to eat and because you are in hurry you don't have time to go through the long menu card, so the waitress makes some suggestion which you can chose from.

10. Hoggington

Posts: 356; Member since: Feb 23, 2016


11. jroc74

Posts: 6023; Member since: Dec 30, 2010

Yea...I agree its on the creepy Pretty soon the jokes about Skynet are gonna be real.

13. SuperAndroidEvo

Posts: 4888; Member since: Apr 15, 2011

hahahahahahahaha Yeah and Google will send you a message to ask why you are not using it. +1

19. ibend

Posts: 6747; Member since: Sep 30, 2014

just wait until it fully evolve.. it will suggest only one option, and its a complete sentence that literally the same as what you think, lol

14. jellmoo

Posts: 2661; Member since: Oct 31, 2011

I think that the flip side is also true though. Just because you sent a picture of your kid (per your example), doesn't mean that the person on the other end either has the time, or maybe even the inclination, for the interaction. This system just helps them essentially move past the notification, while also not ignoring you. Remember, not everyone you interact with is a close, personal friend. There are plenty of looser friendships and acquaintances that people message as well.

29. SmartNingen

Posts: 48; Member since: Apr 02, 2015

Why would you bother sending a loose friend or an acquaintance the picture of your dear kid?

15. jove39

Posts: 2148; Member since: Oct 18, 2011

Hope you aren't mad because Google did this first instead of fruit company! Anyway...other person always have option to type in response instead of using one suggested by bot. I share pics of my kid with family members only and I won't be bothered by their use of automated reply. Coz I know they still love me and my kid!

32. MrElectrifyer

Posts: 3960; Member since: Oct 21, 2014

"Hope you aren't mad because Google did this first instead of fruit company!" Lol, can never rule out that possibility on iphonearena...

16. Valdomero

Posts: 704; Member since: Nov 13, 2012

Instant messaging apps already lack that "human" touch to the conversations, adding now some AI to the mix will make even less "human". I mean, in that case let both AI-minded instant message apps talk to each other about our daily life... creepy...

33. SmartNingen

Posts: 48; Member since: Apr 02, 2015

I doubt the AI will learn my style though. I frequently add *blushes* or *smiles* or *smirks* or *nods head* to try to add that human touch. Or maybe I like Role Play too much. *laughs*

20. sraj49

Posts: 42; Member since: Sep 12, 2014

What was shown was an example. It was also told that the user can switch off of you done want it. Remember what was shown was a preview and jumping to conclusions and writing as an editorial is very immature and unexpected from a mature online publication. Let the product come out, till then you can have your views. This is a free country and you can say what you want or feel, but I am daily raster Phineas and I was disparity worthies editorial which I feel it's immature and paraboloidal.

23. marorun

Posts: 5029; Member since: Mar 30, 2015

Well you know... pro apple ppl tend to act this way toward apple its part of life i guess..

Latest Stories

This copy is for your personal, non-commercial use only. You can order presentation-ready copies for distribution to your colleagues, clients or customers at or use the Reprints & Permissions tool that appears at the bottom of each web page. Visit for samples and additional information.
FCC OKs Cingular's purchase of AT&T Wireless