Welcome to the A.L.I.C.E. AI Foundation

Promoting the adoption and development of Alicebot and AIML free software.

Why You Don't Need Proprietary Bot Software

Noel Bush
June 2001

Noel Bush

The Business Case
(sec. 4 of 4)

Given whatever technological or theoretical argument you may raise, any half-awake vendor of proprietary bot technology will come to its own rescue by promoting the business sense of dealing with a product developed, maintained and supported by a commercial entity. In truth, there is business sense in having an entity that's "behind it all". That's part of why we created the A.L.I.C.E. AI Foundation, to provide a much-needed resource for companies that want to use the Alicebot engine and AIML technologies.

Unfortunately, a commercial entity like one of the proprietary bot vendors doesn't provide a reliable resource. Proprietary software can only be supported to the extent that its owner is able to function. Most bot companies relied on the overhyped technology market of the late nineties to make their start; since the end of that boom, the chips are falling and companies once again have to prove the actual soundness of their business models. Some are making it, some aren't.

What happens when a vendor of proprietary software goes bust? You lose your support. You lose your ability to add new functionality. You basically lose the whole investment you made in license fees, implementation, and professional services from the vendor. Unless you were persistent and smart enough to secure rights to the source code that would at least give you some deeper access to the technology in the event of a financial catastrophe, you are simply out of luck.

All the claims of proprietary algorithms, all the professional services that may be on hand, all the development power that may be exhibited as being available to customers of a proprietary vendor, don't add up to anything except money thrown out the window if that vendor hits hard times.

And what if it's Microsoft that is selling the technology? They aren't yet, but they have a research lab full of brilliant people trying to solve these problems (among others). Would you buy it if it came from Microsoft?

Well, had A.L.I.C.E. never existed, you might. But return to the analogy I started with. Compared to proprietary bot technologies, Alicebot/AIML is at least as advanced, and likely moreso. Since proprietary products are by their nature shrouded in secrecy, and since bot technology especially is something where a lot of smoke and mirrors is necessary to hide the relative lack of differentiation among solutions on offer, making that comparison is, naturally, a matter of some detective work.

You might know that I used to work for one of these proprietary vendors, and you might assume that I'm hinting at some "secret knowledge" about this field--in fact, I'm not. I'm here as a convert from proprietary approaches, telling you that what you can see in Alicebot/AIML stands on its own against any commercial competitors. The fact that A.L.I.C.E. beat out all other competitors, including the product from my former company, at last year's Loebner Prize is a nice testament to the quality of this technology...but as I say, you can see for yourself.

The fact is that even if (all right, when) Microsoft or another giant releases a conversational bot engine, the world today is different from when Microsoft introduced Windows. Back in 1985, Windows 1.01 was released and a scruffy Bill Gates described it to the world as a "revolutionary concept in software"--evidently enough people agreed with him. But there wasn't a better solution already brewing. Apple's new OS was perhaps more elegant in its graphic user interface, but arguably even more closed than Microsoft's, due to its reliance on proprietary hardware. Microsoft was the company that demonstrated to the world--specifically the business world--the need for a business-oriented operating system with a graphical user interface. Even Linus Torvalds's release of the earliest version of the Linux kernel wasn't complemented by a usable GUI for several years. And performance and scalability aside, it's the GUIs like GNOME and KDE that have really begun to break down the doors for Linux in the corporate world.

Today, even a significantly smarter conversational bot engine released under proprietary auspices will have a hard time competing with and staying ahead of the Alicebot engine. Any advance will either be based on academic research, which always comes to light, or on a better arsenal of string-manipulation tricks, which can always be duplicated without looking at one line of source code.

Any proprietary offering will have the added burden of needing to maintain and support its product, whereas a free/open source project receives development support from an unlimited community of interested parties. The ability to support a free/open source technology waxes and wanes with the market demand for the technology. The ability to support a proprietary technology may or may not increase with the financial health of its vendor, and most certainly dies along with its vendor.

In cold hard numbers, a company seeking bot software for its web site, internal network, or whatever other use, simply cannot justify not using a free/open source solution. Paying $50,000, $100,000, $200,000, or millions of dollars simply for a license, and then adding onto that an indefinite need for support from an exclusive service provider, does not a reasonable decision make.

Paying $0 for software, and possibly only incurring the costs for internal expertise to install and customize it, makes a lot of sense. Contracting an independent service vendor (or individual), and having the freedom to change that vendor at any time, is not merely "nice" but the only strategy that makes sense.

Look at the list of companies that are providing Alicebot/AIML services. Today, to be honest, most of these companies are limited in their descriptions of services provided, and limited in available references. Tomorrow that will change, you may be sure. And given that the licensing deals by proprietary vendors and their big-name clients are almost always secretive, you have no way of knowing whether the clients currently listed by those companies (at least those who list clients) are in fact companies that paid a regular fee, or whether they negotiated reduced fees that the vendors accepted in order to "get out the name" (or pay the bills).

The market is simply too young to measure at all. What isn't difficult to see is that even the gap between a publicly-traded company and a couple of individuals operating down in Atlanta isn't as big as one might think, given that the first may have (or may have had) a few millions of dollars to throw around, but the second has at its disposal a technology that has probably had more development effort put into it than all the commercial variants combined.

Last, a few words about "market research". Bot vendors sometimes try to bolster their claims with material adapted from market research companies. There is, for example, a tired old Forrester report that has been re-used over and over again in public presentations and marketing materials by vendors of proprietary technologies, which describes the "cost of customer service" using various traditional methods (phone support center, email), and claims to show how "automation" can save the day. The proposed leap, by the bot vendors, is to suggest that bots can dramatically decrease the cost by eliminating the human operator, or by (in one company's terminology) acting as "tier zero" support. That general possibility is, of course, the premise upon which a recent press release from the A.L.I.C.E. AI Foundation is based ("Bots Will Unemploy You").

But it is indeed a big leap from this kind of speculation, which is certainly relevant but also certainly not proven, and any claim that such costs as are claimed can actually be recovered in this way, using any technology available today. I personally like Richard Wallace's point of view that the "call-center mentality" needs to change before these bots will really find their value.

So I encourage you to look with great suspicion on any report that claims to provide real statistics on the market impact of bots. It is clear that these kinds of technology fascinate people immensely, and that that's their main value in the market right now. Fascination counts for a lot. The kind of "service" they can provide is indeed useful and has value, but is not enough to account for the prices at which proprietary "solutions" are sold.

It should be apparent that AIML/Alicebot provides the best and best-understood technology available for rapidly building a bot that can handle many of the sorts of "tier zero" issues that some proprietary vendors have identified. But it's not at all apparent that that translates into an excuse for selling it or similar technology at movie-budget prices. Unfortunately, most of the market information you're bound to find will attempt to distort that fact, and make claims about the capabilities and value of this kind of technology that are far overstated.

Ladies and gentlemen, please...don't pay a lot for that bot.

<< Previous: The Technology Case | About the Author