Super interesting article about the current state of conversational AI and a tendency for people to be disappointed by its capabilities — the expectation of “intelligence” is not being met. Article features a bunch of problems NLP engineers try to solve these days. Have you heard about the Winograd Schema Challenge? If your answer is “no” - it definitely worth reading the article below.

Understanding what’s possible and what isn’t with today’s AI and machine learning capabilities is key for anyone looking to use such technology to build or enhance applications. If you’re not skeptical enough about claims being made about current capabilities, you could waste enormous amounts of time and money trying to do something that can’t (yet) be done.

https://goo.gl/9XDmCe

If after reading yesterday article you’ve decided to learn more about NLP and want to start with basics - here is a tutorial I’ve recently wrote about NLP engine as a service called api.ai. It was recently acquired by Google and provides a conversational platform for natural language processing and allows us to create bots easily. In the tutorial I’ll show you how to create a simple chatbot which understands basic natural language requests and works inside the Facebook Messenger.

https://goo.gl/Yd2Tiv

Any feedback/questions about the article are highly appreciated in our public chat. And also, thanks for sharing!

Popular voice assistants like Amazon Alexa and Google Assistant have a major drawback: they can only be connected to a single user account at a time. It’s a big problem, just imagine that you can’t prevent your children from streaming audio and video content that may not be age-appropriate. Also, all searches and recommendations through their interactions will be tied to your account and later you can start to receive some Peppa Pig ads after your children searching for a new episode 🐷

But it looks like Amazon is inching closer to a solution because since 2015, it has been working on Voice ID, a way to distinguish between Alexa users’ voices and correctly identify which member of a household is issuing commands. To achieve this, Alexa will attempt to match the voice it hears with a sample called a ‘voice print’. The feature could potentially make it easier to switch between users’ profiles on the fly and also to set certain functions – like placing orders on Amazon or running home appliances – to require authorization from a specific user.

It’s likely that this feature will gain a huge advantage in the game, as they’ll not only be able to promise more convenience, but also enhanced privacy controls for things like banking, as well as more personalized responses.

https://goo.gl/OfVK9P

In addition to the last post about Voice ID: Amazon debuted a Door Lock API that gives the makers of locks connected to the internet the power to lock or unlock doors with Alexa. The API is part of the Alexa Skills Kit.

“We’re currently working on a variety of security and safety tools within the Door Lock API in order to support unlock controls in a secure manner. We will not launch unlock support until those security requirements and options are in place,” said an Amazon spokesperson.

And again: ability to combine this feature with the Voice ID (distinguish between voices) could improve security for devices that use the Door Lock API to work like a charm🔥

https://goo.gl/OWiZwK

2016 showed us that your bot doesn’t work out if the conversation UX of your bot is bad. That’s why I want to post more about the good examples of conversational UX.

One more great article which was born after analyzing a plenty of user’s conversations with bot. It reveals 2 specific problems users face when they, let’s say want to book a room for 3 nights in San Francisco on Monday:
1. The first mistake developers or product people make is to process each input as a command to trigger an action from the bot. Usually users don’t write one big sentence like “I want to book a room for 3 nights in San Francisco on Monday”, their mind flow produces 3 messages: “I want to book a room”, “I’m in San Francisco, btw”, “I’m thinking about 3 nights on Monday”.
2. The second mistake is not to process typos. Users often do something like: “I want to book on Dec 28”, then “oh shit I meant Dec 27, sorry”. We need to care about it too.

Elegant ways to solve these problems is in the article below.

https://goo.gl/1o1z9J

Facebook Messenger has just released a HUUUGE update! I cannot wait to share it with you. Ton of cool new features like persistent menu localization and programmatic way to retrieve Messenger code. Awesome! 🚀 🤖

Messenger Platform Changelog — https://goo.gl/XTibZv

P.S. As you noticed, I found a nice way to share the content I generate here to Facebook & Twitter, with nice preview and link to the post. Hope you like that!

Interesting point of view in the article from TechCrunch reviewing the latest Facebook Messenger update which was released yesterday. The biggest update was done for the thing called “persistent menu” - it’s a navigation menu you can always access from any point of your interaction with a bot. So, this menu was rebuilt - you can now add categories/subcategories to it. It was also brought upfront - so it’ll most likely be a starting point of user bot journey. And the most important thing here - they have even allowed to hide the text input field in order to turn off all the text conversations. And I like it the most - right now you can decide which one bot to build: smart chatty assistant with great NLP capabilities or lightweight buttons/menus app inside messenger. One more step into a WeChat story.

https://goo.gl/W2dgFu

Yesterday, while surfing through the internet I came across an article from Typeform about the rise of the conversational commerce. I started reading it and after the first paragraph some weird thing has happened on the web page. A couple of seconds of frustration. But then, when I understood the concept of a reading process, my mind was blown. It’s so simple and so engaging at the same time. I don’t want to spoiler, so just go and try to read this article. I promise you won’t regret.

https://goo.gl/9m2vG9

Hey folks, I’ve finished my 5-days work on a huge compilation of different resources related to bots. It’s kinda big and includes everything from tools for development, marketing, analytics for bots to the events, books, influencers and newsletters. I hope it will be helpful for many newcomers and can bring a lot of value to the community :)

I’ve decided to put it on Github so feel free to share it with your friends by clicking ‘star’ button and also create new pull requests if you think that something is missing. I really appreciate your help, thanks! 💪

https://github.com/BotCube/awesome-bots

What does Sunday mean for you? For us it’s the #botoftheweek nomination at BotCube of course 😍

How do you have better meetings? Create an agenda. Share it before the meeting. Ask better questions. Take notes. Write down clear next steps. None of this is rocket science. The hard part is finding the time to actually do it. That’s why I want to show you a GoodTalk. GoodTalk is a a slack bot that helps make your 1 on 1s more productive, fun, and engaging 🎉

GoodTalk provides managers with curated questions and best practices so you can be more prepared in less time. GoodTalk prompts managers and employees to add things to the agenda between meetings. No more awkward silences. GoodTalk creates a dedicated space to build stronger relationships with your employees by showing progress. Staying on top of the little things makes a big difference.

https://goo.gl/jT0EBR

Facebook is testing a new reactions feature in Messenger, which would allow users to add reaction emojis next to text in conversations. Users would do this by clicking on a button next to text, which would bring up a series of reactions. The reactions are similar to the current six reaction buttons for posts, but also includes a thumbs down button 👎 (yep, dislike).

The feature also includes a reaction counter, which would list the number of reactions on any given line. The feature doesn’t appear for all users, but the company has conducted tests on small groups of users before rolling it out to the larger population. I hope this feature will also be accessible through Bots API, will be a great replacement of a custom implementation with quick reply buttons.

http://www.theverge.com/2017/3/5/14821874/facebook-testing-reactions-feature-messenger-dislike-button

Do you remember my post how it’s going with AI in Apple? Here is a new one from the same author (Steven Levy) called “Inside Facebook’s AI Machine”.

Facebook is relying a lot on AI. It cannot exist without AI today. Every time you use Facebook or Instagram or Messenger, you may not realize it, but your experiences are being powered by AI. Even Zuckerberg’s 5700-word manifesto invoked the words ‘artificial intelligence’ or ‘AI’ seven times.

If it’s too long for you to read - don’t skip, just add it to your Pocket to read later. By the way, thanks @flreln for this article.

https://backchannel.com/inside-facebooks-ai-machine-7a869b922ea7

About AI in Apple: https://news.1rj.ru/str/botcube/48

Do you remember that story about Amazon Echo and a murder case? If not - the link is below.

Finally, Amazon has agreed to hand over data from an Amazon Echo that may have been operating as an alleged murder took place, after the defendant consented.

A hearing is scheduled for Wednesday.

http://www.bbc.com/news/technology-39191056

More about the case is here: https://news.1rj.ru/str/botcube/62

This is an early stage of the moment we have all been waiting for: WhatsApp reportedly started testing system that would let businesses talk directly to WhatsApp users. Seems like we’ll see such a desired bot platform API inside WhatsApp soon (in half of year 😂). I’m happy because our clients have always asked us to build a bot inside WhatsApp but it wasn’t possible in a native way.

Also, it’s an important signal of how WhatsApp plans to make money from its massively popular service since it has not developed a business model in the three years since Facebook Inc bought it for a hefty $19 billion.

http://venturebeat.com/2017/03/08/whatsapp-reportedly-testing-business-chat-tools/

I’ve just finished reading a very long article about one interesting experiment which was done with a Kik messenger chatbot named Kokobot. But the main story here is not about chatbots at all, it’s about people.

Koko is ushering in a new era of emotional well-being, providing emotional support as a service to any product. With KokoBot, users have access to a peer-to-peer network that helps them deal with symptoms related to stress, anxiety and depression. For people who are suffering, knowing that others are in pain, and that they can do something about it, is one way of healing themselves.

I highly encourage you to find some time to read an article below about the story how chatbots can help people with mental diseases.

http://www.theverge.com/2017/3/2/14764620/koko-social-network-mental-health-depression-app-kik

Try KokoBot on Telegram: @thekokobot

Nowadays kids adore their new robot siblings which help them turn off the lights, answer homework questions and entertain friends at sleepover parties. While helping kids and bringing value to them these assistants can also be harmful in many ways.

Some parents think that Alexa tolerates poor manners. To ask her a question, all you need to do is say her name, followed by the query. No “please.” And no “thank you” before asking a follow-up.

Also, it creates patterns and reinforcement that so long as your diction is good, you can get what you want without niceties.

Parents have noticed that queries previously made to adults are shifting to assistants, particularly for homework — spelling words, simple math, historical facts. Kids will go to their parents less, with both sides losing out on timeworn interactions.

It’s a very hard question though to discuss in a couple of paragraphs, so I encourage you to read the article below to understand all the implications behind kids interacting with the robots.

https://goo.gl/4FWIEe

I think that those of you who use Facebook Messenger on a daily basis have probably noticed that not long time ago it was sufficiently redesigned to add more camera features upfront. And as I’ve noticed most of my friends are completely annoyed by this change. So, what’s the implication behind this? How Facebook Messenger transformed from a chat app to a chat camera? Here is a great article from TechCrunch explaining the historical part of this phenomenon.

https://techcrunch.com/2017/03/08/the-camera-is-the-keyboard/

It’s Sunday, so #botoftheweek is here, guys. Today we have a very interesting and useful one. How many times have you been suffering trying to come up with a new project name and then checking all the domains and social account availability? For me it was a real problem though.

So, here is a bot called Namely. It’s a Facebook Messenger bot to help you decide your next projects’ name. You send it a name proposal and then it checks all the domains and social accounts availability for this name. Sounds like a huge time saver, right? 🙏🏻

https://www.messenger.com/t/416101528740470

We have an article from a non-technical source today (Wall Street Journal) about AI. Basically it’s an interview with one of the most popular influencers in the world of AI nowadays - Andrew Ng. He is a chief scientist at Chinese internet giant Baidu Inc. and co-founder of Coursera. And yep, he’s also a teacher at the super popular “Machine Learning” course at Coursera. They talked about speech recognition, singularity, automated jobs and plenty other hot topics. The main idea of the discussion though is that the hype about artificial intelligence is real this time.

https://www.wsj.com/articles/how-artificial-intelligence-will-change-everything-1488856320

It’s always a pleasure for me to post here articles about Slack team and their updates. I think that’s because these guys are awesome and doing really well on constantly improving and engaging a communication inside teams.

And here is a new update - now they’ve improved a way we’re sharing and previewing links. Basically right now with a new SDK you can show some authenticated content at the link preview and also add some buttons to it to bring interactivity. More info is inside the article, these guys are rocking 🤘🏻

https://medium.com/slack-developer-blog/all-will-be-revealed-ebcad7c531f0

Capital One has announced Eno - the first natural language SMS chatbot from a U.S. bank, allowing customers to ask questions using natural language. Customers can ask Eno questions about their checking and savings accounts using a text interface. Examples might include “What’s my balance?” or “How much credit do I have?” 🤑

The idea of this chatbot came up when the bank found that when it sent out automated fraud alerts through SMS, customers thought that the sender was human and started interacting with it.

This is remarkably different than other brands chatbots on platforms like Facebook Messenger and Skype. Capital One believed Facebook has too much access to peoples’ private information, any Facebook Page admin can view the personal banking information of customers.

https://www.macobserver.com/analysis/capital-one-natural-language-chatbot-eno/