Ask HN: Your opinions about bots? Facebook, Microsoft are putting lot of effort behind pushing bots inside a conventional ui. What's your idea about bots? Worth investing? |
Ask HN: Your opinions about bots? Facebook, Microsoft are putting lot of effort behind pushing bots inside a conventional ui. What's your idea about bots? Worth investing? |
Some examples I see is something like
- "hey $financialbotnam"
- "Hey $user what do you want?"
- "Please send me my transations for groceries, bla and bla for july 24 2032 until august 23 2064, thanks"
- "Okay!"
- "Your transactions are bla and bla"
- "Thanks $financialbot, and what about my blah and bleh?"
- "bla bla bla"
And so on, seems just.. annoying.
That being said, on my local phone-website there have been one of those "Ask Sara" bots (since like 10 years) which I found works okay for what it is, but I suspect (and treat it like) it is just some search. So I usually just ask some keywords.
Which is kind of my point against the bots, I don't want to type stuff..
And I don't use voice stuff like for Siri etc. But if I did I might have a different opinion.
One type of chatbots that I think will be revolutionary is making chatbots that bring people together. Messenger just hit 1 billion users[0] so that's a huge userbase that's already connected together just by virtue of having a Facebook account. Chatbots imo have the advantage of not needing a "share with your friends" button that apps do. For chatbots, sharing and social connectedness is already built into the platform and the winners will be the ones that take full advantage of this.
Of course I might be biased because I am currently building a chatbot and have made one that is slowly gaining traction. But this is just my opinon, and the verdict is still out on whether this is just a fad or here to stay.
[0] http://www.theverge.com/2016/7/20/12235476/facebook-messenge...
At this point I cannot say to Cortana, "the Camera App doesn't work, it probably requires some drivers, fix it" and get any meaningful help out of it.
AI is a bit of a counter-intuitive market: it's a high growth segment, but the structure of the market top heavy, and several things make it an unattractive investment:
1. Democratization: much of the technology is the public domain, meaning anyone with a CS degree can muddle together their own (and many have). That means if you invest, you investing in pure AI, which means you're investing in cutting-edge technology - i.e. Research, which very difficult to assess.
2. Dependencies & Go-To-Market: intelligence (i.e. pattern predictions) that's actually useful on a market wide basis require massive amounts of training data. There are a couple of resources, like ImageNet & Yahoo's data dump, but in the interim - AI needs to be domain specific. As such, the dependency is a widely used application (i.e. Facebook, Instagram, Google). You see the dilemma. Therefore, what seems the likely outcome is acquisition.
3. Unit of Economics: For this sector, it's the accuracy of its predictions. This is almost too one dimensional to assess as the basis for investment. Acquisitions decisions are made almost exclusively by the R&D department. Therefore, if you invest, you need to know what makes one AI different/better than the next.
IMO, it's too early to put money in. The structure of the market is still nascent, heavily favoring incumbents (Google & Amazon offering AI predictions as a service). It's difficult to see how an AI startup can break in, and how it can be profitable.
Take the above with a grain of salt, as it's tailored to my risk appetite.
The thing is, if a company is actually able to figure out how to do this conversational AI right, that company is be a game changer.
Bots emerging from finance, law, academia, eastern european kids in a garage? Treasure.