It’s been a busy week for chatbots, not surprising though as Apps have now become the big lake/mino effect and everyone has realised there’s probably not enough money in the “App Economy”.
How Much Interest?
Enough that it looks pretty obvious in Google Trends.
Now IRC and AIM Messenger chatbots have been around for years, I was writing AIM bots to do common tasks (Weather, RSS Feed News and Flight info) back in 2002 at the same time I was getting the data mining bug. The one thing we didn’t have then was someone else hosting the platform, it was hosted and run by the owner.
Now we have the added fairy dust of AI, high volume data processing and a tech press and investor structure who’ll throw money at the next big thing.
The issue is that the deployment and hosting landscape has changed considerably. With Facebook and Microsoft firmly putting their flags up in the bot arena it’s the big companies that are offering the tools to make your next fortune (in theory). Those frameworks come at a cost, data privacy.
Who Owns The Conversation?
This will be the most important question. And where you host the chatbot of your dreams is going to be pretty important. As with apps the main challenge is going to be the distribution of your chatbot. A chatbot is just another app, people will line up with their own app stores. Telegram already have an API to integrate chatbots.
Training the AI is hard if you don’t have the experience, probably why I’ve had a few enquiries over the last week. I don’t invest, I don’t want to be your tech co founder (that old chestnut) and I’m not going to build it for you. There’ll be a host of accelerators popping up to get your chatbot to market. The majority will dodge the most important question, who owns the conversation, as it’ll be about being the first out of the door and investor ready.
Putting your product on someone else’s platform is always a brush with danger, regardless of what you may think, you’re losing control of the core platform. While I agree that the larger companies will have already done the training involved there’s the risk of that Nest moment where your bot is shuttered for no reason whatsoever. Lest we forget the big Twitter plan on the API cherry pick back in 2012?
Azeem Azhar nailed in his excellent The Exponential View newsletter, “There are many reasons to follow the new KISS. (Keep it Separate, Stupid).” If you’re not reading it, then subscribe here, it’s become my essential Sunday morning reading.
Is it really AI?
Over time people will realise that doing AI is hard. The problem right now is the transition phase where everyone is reading about AI in the media and the explosion of startups waiting in the wings. Remember that annoying quote about BigData and sex? Just rinse and repeat that with AI chatbots.
“An AI chatbot is like teenage sex: everyone talks about it, nobody really knows how to do it, everyone thinks everyone else is doing it, so everyone claims they are doing it…”
As I guessed a few weeks ago there’s a good chance that you just need to expect the obvious and realise where there’s an “AI” thing, there’s more than likely a human somewhere along the line. Some of the time it’s to keep training the algorithm and that’s fine (and needed). What I believe you’ll see is the rise of companies and entrepreneurs touting AI and just having a human responding to a Web/Phone/Text request.
Lest we forget the sorry tale of Spinvox who’s Voice Message Conversion System (VMCS) was sending voicemails off for human conversion, usually outside the EU, which landed them in a lot of bother.
The Coming wave….
So, the coming wave of AI startups – a mixture of technology, hype (and let’s be honest, it’ll be mainly hype) and human interaction. It’s like the App explosion of 2007/08. No one will be making $400k from a farting AI though.