Fallbacks are not Conversational
Story telling has risen to be one of the holey grails of digital.
Folks have figured out that providing a tool for people to express themselves is a goldmine — ‘cause all they have to do is just step back — and let users “do their thang” — and the tool/technology platform vendor — benefits. Directly. With uptake power and influence, which can be marshaled and exploited in numerous ways.
Monetizing that benefit is the tricky part — especially when you’re a platform vendor with hopes of bringing OTHER tool and service vendors onto your platform. Who’s paying for what? What % of the pie do you (the vendor) demand?
To me (the most important question) — is what I live for everyday: “what are the capabilities, toolkit and built-in functionality that a vendor provides — which I can THEN build on top of?”
I am what’s called a “platform developer.” I build on TOP of other’s platforms. I been doing that since 1984 and the advent of the Macintosh platform.
In fact we were developer #10 in Guy Kawasaki’s Army of Mac developers — who took on the computing world — and won.
The world of ChatBot and Messaging platforms we find ourselves in today— is currently being led by Facebook — with the triple threat of Messenger, Instagram and WhatsApp.
If you look at it from not just Messaging POV, but in terms of creative usage of video and on-line Video — it’s YouTube, SnapChat and Instagram — which are the three powerhouses.
What excites the SHIT OF ME is what happens when short form videos — and Bots — intersect with Messaging and embedded UI’s.
But I digress………… — — — — — — — — — -………………………
Lets not worry about what I’m working on — and focus on Facebook Messenger for this article’s sake, their new Day feature and the current set of issues confronting Facebook ChatBot developers.
Facebook has become experts at pushing developers buttons and offering them features that then motivate developers to work their asses off, mortgage the family farm and devote 80–100 hours a week — for that brass ring in the sky.
Which leads us to a world where ChatBot platforms are perceived to be the ticket to the next gold rush, technology platform = unless you believe Robert Scoble.
Fortunes have been built on Apps and platform features — which TODAY no longer make sense as stand alone products. But at the time those stand alone features — were an entire website or App.
We’ve all witnessed “being in the right place at the right time” — and we’ve all seen these features become intrinsic part of larger platforms. A single technology feature can sometimes be sold for $100,000,000's and swallowed up in the bowels of some tech behemoth’s platform — without even a “burp!”
THIS has led to a climate and startup environment of 50,000 Don Quixotes charging at Windmills of all shapes and sizes — running on the belief that if they can be there FIRST — they’ll sell out to Facebook, Google, Apple or….
Facebook is partially responsible for creating this kind of environment (along with Google, Apple, Amazon, Microsoft, SAMSUNG, etc.) and now WE need to be smart enough to see through the economic motivation — and keep our eye clear on the prize.
A new form of self expression.
We’re DAM CLOSE!
Facebook is faced with a predicament. They’ve unleashed yet another gold rush around ChatBots and now they’ll have to deliver on the promises they hand out at their yearly f8 Developer conference.
Boo Hoo Hoo poor Facebook! Actually has to DELIVER on their promises!
Early results are in — and they ain’t pretty.
Surprise, surprise its not that easy to just develop “a bit” of AI ChatBot platform technology and “sell it to Facebook.” I just believe that 25,000+ ChatBot developers — producing 50,000+ ChatBots — can’t be THAT stupid.
Its greed that’s driving them, not product or change.
These greedy 50,000 land-grab experiments (which yield results like 70% of all human responses go misunderstood) are the results of Facebook “pushing” ChatBots extremely hard. Now it looks like they’re changing their “focus.”
With an upcoming f8 quickly approaching Facebook has begun to roll out what “some” of their next playbook looks like.
Just in time for SxSw — we’ve already seen the release of Messenger Day — a SnapChat story clone and the release of Version 1.4 — which saw these “exposed Menus” at the bottom — instead of “hamburger menus.”
I’m still grappling with how to use Day — but early feedback says “this sucks.”
V1.4 and Day are just the warm ups so that all of f8’s announcements don’t overwhelm everyone — all at once. Nothing wrong with giving developers a head start, but its extremely obvious that this is all part of a concentrated re-positioning campaign.
Don’t you just LOVE feeling manipulated?
I love the idea of building story telling into Facebook Messenger, but I was really hoping that Facebook would have put Bots into Instagram rather than InstaStories into Messenger.
That’s where I’m coming from — but I understand why Facebook did what it did.
They see Instagram/Snap story traffic driving growth — and they want Messenger to do — everything.
My daughter will never use Facebook, but she’ll use Instagram. I will never learn Snap, but I yearn for a Day feature — which gives me the full range of creativity and expression in Snap.
I just don’t like the temporal nature of it all — and :10-:15 sec. video lengths remind me a LOT of 140 characters.
That’s just me.
— — — — — — -Now back to my point — — — — — — -
Because Facebook has launched one of those generational tsunamis techies live for — AI —they (Facebook) now have to live with the ramifications of their greed, over reaching and Wall St drive growth obsession.
“How does Facebook promote their own agenda, keep Wall St. happy (by staying profitable) and save the world‘s problems— while delivering quality ChatBot AI capabilities?”
Their answer (in typical Zuck style) is to “hedge their bets.”
Facebook has been developing a hybrid Bot/Human technology called “M” for years now. I assume it’ll be a major aspect of Facebook’s f8 focus on business. Facebook has to show Wall St that it’s meteoric growth will continue to be fueled by tackling the Business World. But I’m not here to worry about M or Facebook’s business aspirations.
Yes — developing a platform explicitly for humans AND Bots is kinda awesome.
And Yes — if I cared at all about enterprise, corporate issues or business issues (which I don’t) — I would ramble on for DAYS about M and the next gen of AI in business. I could have a consulting company, start a conference and get all “leadership” and shit about all this. But I don’t and I won’t.
So THERE!
What I care about are humans, independents and creativity and enabling those humans to express themselves.
That’s Messenger and wit.ai and all the ChatBotty shit they been shoving down our throats over the past year. This is a completely different story than M — and that’s partially where the confusion lies. Certainly none of the press get that.
The relationship between humans and Bots is completely different than a “purely Bot” (virtual asst) kind of relationship. Facebook is trying to do BOTH!
So in the spirit of keeping track of all the issues, this Josh Barkin article got me considering an overview of the issues that we are currently faced with.
In no particular order I’d like to use this opportunity of reacting to each of these current issues — as each topic is a world unto itself — yet Bot developers have to deal with ALL OF THEM simultaneously.
Think of this list as the TOC for the entire Facebook ChatBot developer universe:
- Standards
- Data
- Embedded Apps and Conversational Interfaces
- Conversational Messaging
- AI — and NLP
- Discoverability — which I’m purposefully leaving for last
I’ve tried to “jump all over” Facebook’s recently released “Day” (Snap story feature clone) feature and this has led to to assume that there’s NO WAY IN HELL Facebook thinks this version of Day (where it rudely inserts itself into the Messenger UI) is the final version of Day.
its just too raw, has too many gaping holes and is a bit too obscure for any kind of uptake. So consider yourself all test monkeys — people.
I believe that the same discoverability challenge Facebook faces with 50,000->500,000 Bots is in fact — the SAME discoverability issue it faces with Day stories.
But that’s Discoverability — see below.
Lets start off with Standards — one of those holey grails — that keep our imaginations alive and burning.
STANDARDS
I’ve lived my life helping to build, establish and grow standards.
The very nature of Trivia Blast utilizing a Messaging App to play games —by providing cute little buttons to tap on — is a standard.
They way that Sensay and OlaBot delve into one’s personal life and reduce it to one of five choices underneath the message — is a standard.
The way untold 1,000s of Bots render their catalogs and range of choices — into horizontal scrolling Menus — is a standard.
There are plenty of standards in the “conversational interfaces” possible with ChatBots. But do NOT expect there to be any sort of interchange or import/export of content, experiences, users or data.
That ain’t gonna happen.
Its up to us to establish our own way of “normalizing” across the various versions of our product. Which leads us to “the data.”
DATA
Your AI is only as good as its data.
What’s missing in most ChatBots today is any semblance of the importance of built-in data, which (by definition) would enrich and improve the user’s experience with the ChatBot.
One rule of thumb: you’re gonna need ~10,000 responses, topics, sentences, comebacks, etc. built in — just to say “Hello World” — and another 10,000 items which are “domain specific” to launch an MVP. If you have less than that — expect user failure.
One way to think of it — is that successful bot vendors of the future will take an attitude of building data INTO a tool.
Think of it as stock libraries (Getty=Images, VideoBlocks=video, SketchFab=3D, Thingiverse=IoI items) of conversational questions and answers.
Think about building in databases of (what’s called) labeled data: you can find this stuff at: Cornell Univ’s Movie Dialogs corpus, Paul Allen’s “Anything on Everything” (Aristo Tuple KB), Freebase (now archived.) eg. follow the data.
Think of it as huge databases of templates that handle self-configuration or customization of “floor tiles” or “plant arrangement” (or whatever the hell domain you’re specializing in.)
AI means lots and lots of data — or else its gonna fail.
Embedded Apps and Conversational Interfaces
Please don’t let me get pissed off. This is what Cola was all about. But it got sold for $75,000 (to Layer — so congrats to THEM! They got a great deal!) I really believed that the patents alone would have justified a higher valuation, but that’s life folks.
Cola is a platform for developers to build their OWN embedded Apps inside of a Messaging Flow. WeChat calls them “mini-Apps.”
Embedded Apps (and UIs) inside of Messaging flow is a key next level of functionality and end-user capability in the move towards “Messaging is the New OS.”
The natural extension of that is when you add ChatBots into the Messaging flow — and now you have your AI stampede and greed.
Whether it be an entire ChatBot system, videogame or real-time therapy session — embedded Messaging Apps mean you can add ANYTHING into the flow of a Conversation. With Cola — we hoped to make THAT the ecosystem.
Oh well. But it did lay the groundwork for what comes next.
Be prepared for a different kind of Messaging experience.
Yes — you can create a Bot to answer questions — but you can ALSO present any OTHER kinds of Interactive Interfaces to ChatBot customers as well!
It doesn’t have to be JUST the UI toolbox calls that Facebook provides!
Webviews can be seamlessly integrated into any ChatBot!
Conversational Messaging
Which leads us to the very notion of Conversational Messaging unto itself.
It would seem to me that to create a conversation — it has to be two-way.
So how come so many ChatBots today are one-directional?
What do you THINK users (customers) would say — when they’re greeted with the same dam response — over and over again!
Thus — the title of this article: “Fallbacks are not conversational!”
AI — and NLP
For ChatBots to be successful they gotta respond.
Doh!
Not much more I can say about that, except that AI is a great buzzword, it attracts investors and futurists, but it costs money to be a playa — YO!
So don’t be talking shit, claiming you’re doing AI when you AIN’T homeboy!
And homegrrrrrrrl.
Don’t be giving long, emotional speeches about dead colleagues and capturing the essence of their being — inside a Bot — when in fact — you haven’t.
‘Cause people are gonna try out replika.ai — and find — nothing there.
Discoverability — which I’m purposefully leaving for last…..
This all leads me to finding all these Bots.
How the HELL can we expect an ecosystem to grow when you can’t find the dam Bots in the first place?
Everyone is speculating that a Bot Store will be released (or at least discussed) at the upcoming f8 developer’s conference.
I certainly hope so — myself.
In fact “a little birdy told me” that ONE of the value added aspects of Viv (which was sold to Samsung) was its approach to providing a Bot store.
But since Viv has not shipped yet, we’ll have to wait on any details of that rumor.
In the mean time — I hope you’ve gotten some value out of this article.
Canter — out.