Very general question, but I am interested in all kinds of opinions. From my current understanding there are basically two types of uses for chatbots:
Entertainment (“chatting”)
Informational (find an answer to a specific question).

So where are these going to go. There are high cost commercial vendors, low cost more technical orientated providers, people just creating their own bots and extending their human-likeness, open source development of new engines etc.

I see the future for chatbots as being included in household objects like TV sets and toasters, where you can simply talk to your goods and they will react accordingly. Just imagine being able to ask your TV, “What time is the football on?”, “Can you record the news at 10 please?” or “Is Star Trek on any channel soon?” and the TV replies back to you.

We have already seen it in phones with SIRI and the various Android versions and I genuinely see this becoming part of our daily lives.

A good market would be for talking toys as well. Putting a chatbot inside a teddy bear or doll, would be a hugely popular toy.

A couple of threads that relate to your question:
http://www.chatbots.org/ai_zone/viewthread/776/
http://www.chatbots.org/ai_zone/viewthread/673/

Intelligent Agents will be pervasive in your appliances/car/mobile tech and acting as a personal assistant in your daily life.

Douglas Adams covered it pretty well in “The Hitchhiker’s Guide to the Galaxy” where he wrote about “your plastic pal who’s fun to be with” and even the automatic doors insisted on having a conversation with you as they went about the business of opening and closing. Heaven forbid it should come to that.

So is the Future of chatbots just in the control of devices?
Are all the entertainment bots, that are placed in the challenges just going to end up in some toys?

And where do you see the future of the fundamental technologies? Are ALICE based bots with their “simple” pattern matching going to be replaced by something more sophisticated or is there still a lot to go with that technology?

Where are the big stumbling blocks for the future of chatbots, what limitations to crack?

I would say the biggest stumbling block is lack of funding. There are a number of smart people with interesting ideas, but very few that are able to produce a revenue stream that can help accelerate the R&D work. As such, chatbot tech takes a “hobbiest” approach. It is commonly developed by individuals as pet projects.

One example is the Zabaware Open Source 3D Character Engine, they tried a Peerbackers fund raising drive to support an open source project developing the next generation of Ultra Hal and 3D character animation technology for the greater chatbot community. Fundraising has been meager.

Bora - Jun 13, 2013:

Where are the big stumbling blocks for the future of chatbots, what limitations to crack?

My opinion is that chatbots are just another AI application, and that all AI applications have the same stumbling block: we don’t yet know the most important fundamentals about how biological brains work, especially the form of high-level knowledge representation used in brains. Once that relatively simple question is answered, progress in AI should be *very* rapid, and chatbots will ride the wave of progress along with every other AI application.

I see there are a lot of vendors already in the Virtual Agent space and larger software companies are starting to buy them up. Certainly an early adopter market, but it should move to mainstream into the next years. I am interested to see how much sophisticated NLP technology will bring a key differentiation for these vendors or if it is other factors.

Entertainment bots are another thing. Simple ones for toys should come out more I would expect, but I wonder how it will develop for e.g. Ingame bots (hardly see them) or your home appliance with a personality. Is that the next hit or do people not want their systems to have a personality?

From what I have seen until now it is still a lot of work to create a bot that one can have a conversation with. A lot of technical knowledge as the programming isn’t on a non-techie level, nor seem there be many easy adoptable / pre packaged response lists. Don’t get me wrong, I know that a lot of bot masters provide their scripts for free or you can e.g. Buy superbot, but “quickly” reviewing or editing 10,000+ responses sounds very tedious. This requires a lot of passion / potential benefits / pressure before a company will do that.

But all that are just my 2 cents and I might be wrong.

One “serious” use for chatbot that comes to mind is something Dr. Wallace would sometimes mention at various ALICE/Pandorabot web pages.  I recall reading of how a chatbot could be tailored to respond… not only with what you might say, but how you would say it.  This could afford future descendants the ability to know you in a way that’s not currently available from simply looking at old photos or records.

I’m interested in genealogy, so the idea appeals to me.  I can also see a business opportunity for someone willing to invest the time.  And as the future brings about improvements in animated faces and voice capture, it’s not difficult to imagine how something like this would be better than simply watching a video of a deceased relative. 

Conversing with someone who died before you were born would be an interesting experience.  My grandfather would often tell me of how different the weather used to be, and he had a few favorite expressions.  When he’d get angry, rather than just swearing, he’d say, “Jesus, Mary, and Joseph.”  I can hear his voice and see his face as he would say things like that, but I can’t share them with my children who never knew him.

A second thing is what’s being experimented with now regarding areas of the heath care field.  Chatting with a pseudo-person might be preferable to filling out forms regarding someone’s current health, as well as diagnosing an underlying illness.

I’ve actually had several people approach me with questions about creating “virtual copies” of themselves to give them a sense of immortality, and to help future generations understand more about them. One even wanted to have the chatbot able to advise a granddaughter on whether or not to “marry Billy”. I think a project of this sort would be VERY involved, but it’s certainly possible. However, to create a basic, primitive version of oneself, the hours involved with creating the nearly half million or more AIML categories required would quickly add up.

I’ll second Steve’s response about the talking toy/companion:
[ul][li]http://www.youtube.com/watch?v=3JMvs5f0Mks[/li][/ul]
I was approached by a group that wanted to create a companion but turned down the work as I was overwhelmed and the company that approached me did not want to do the animatronics.  The company also expected me to have the knowledge to embed the technology on a chip and did not want to invest in the additional engineering expenses necessary to make this a go.  There is definitely a market out there with people willing to pay $‘s to talk to a A.I.  My chatlogs on my paid android application show a healthy percentage of purchasers into their 30’s and 40’s, so this market is bigger than children toys.

Also likely at some point in the near future are storefronts and websites that virtually interact in an informational/selling mode once you enter their premises or visit their website.  In a physical store, you could even be “tagged” with facial recognition and the virtual sales agent could then “follow” you around the store with speakers and mics at various access points within the physical store. We’ve already seen this conceptualized in the movies.

And of course, at some point in the near future, we all will likely have a personal attendant A.I. that helps us manage our lives.  We are well on our way of meeting that goal.

Brian - That’s the sort of thing I would love to see Mitsuku in. I agree that whoever develops something like that would make a fortune. With Program AB being able running AIML without a server, I am sure some bright spark could put it on a chip and possibly have a USB port in the toy to allow daily or weekly updates.

As for AI Teddy Bears, long ago my kids had a Teddy Ruxpin.  It did some clever things while kids would read a book with them.  A while back, there was one or more toys that would interact with a TV program, which was kinda scary.  Think of the possibilities.

As for the “personal copy,” I think Pandorabots is well on their way for something like that.  I’d imagine a person could fill out a form online with some basic information, and then, with a bit of basic instruction, go more into whatever subjects might interest them.  I’m convinced that a lot of ALICE/Pandorabots is actually Dr. Wallace’s preferences, opinions, and philosophy.  And as technology advances, I think maybe it can all be accomplished with something like wearing a headset.

I personally prefer machines to be machines, and if they talk and/or listen to me, I’d expect them to do so with as little fake emotion as possible.


As to making a copy of one’s mind, cl0ne (http://www.chatbots.org/chatterbot/cl0ne/) seems to be an attempt to clone a person’s personality by chatting with it.


I’ve heard in Japan they’re already building robots for caring for the elderly, and they want those robots not only to be able to communicate via spoken language, but having human personality traits, too. (A first step into a Douglas Adamsian universe, Andrew?)


But if you want to build a machine that behaves like you are used to (a “behavioral clone”), I’d recommend to ask Google to do that, using all data that exist about you; best would be if you’d record every single of your tiniest muscle movements, words, and changes of electric potential in your brain.

I know Peter has a lot more experience talking to bots than I do but I could never get any sense out of cl0ne.

Also, if or when I get into my senior years, I would hope to have some human company rather than have a robot care for me.

In spite of logical preferences, the numbers don’t lie.  A growing elderly population, as well as a shrinking birth rate, are going to require some creative solutions.  We already employ monkeys to pick up dropped items, to open cans, and to get a beer from the refrigerator.  Dogs are used in various ways to assist people who can’t see, can’t hear, or to calm those about to experience an epileptic seizure.

I don’t think there’s anything wrong with “artificial” assistants, as long as we remain cognizant of the fact that they’re not living beings with self awareness… however, it’s not always easy to suspend disbelief.

http://www.youtube.com/watch?v=zCleX01rPqI