-Under Construction-People have tried to get an AI to have a reasonably intelligent conversation for a long time. Cleverbot doesn't pass a Turing test for long, but it certainly is amusing...
People always dreamed about being able to talk to machines, though not all of them believed eating a red pill was necessary in this matter. It turns out that they were right, but the red pill kind of helps anyway.
In 1964, computer programming pioneer named Joseph Weizenbaum started scripting a computer program that could communicate in standard English with a human. This took two years to complete, and when it was ready, Weizenbaum dragged the 50 pound hard drive up to the roof, into a severe rainstorm. As lightning struck, he shouted "It's alive! ALIVE!!!", at which point the program asked "What do you want it to alive?".
Weizenbaum named the program ELIZA, and tried to teach it to stop being such a dumb fuck all the time. It was just impossible though. ELIZA could "make conversation" by using pieces of user input, but ultimately it didn't hang together enough to be believable. AI scientists use a simple test, the "Turing Test", to find out if an artificial entity has reached a human level of verbal intelligence. If the machine can talk to a human so well it passes for human, then it passes the test. If the human can detect that the machine is a machine, it fails. It's too bad all scientific tests aren't that simple and straightforward.
One interesting and unexpected result from all of this was the computer text adventure. Almost all of the knowledge and theory gained from ELIZA was used by college students to program text games that soon evolved into "Zork". The popularity of these games did not last in comparison to videogames with graphics, but they never went away either. Even after Infocom went out of business, plenty of people kept writing games like this themselves.
Since that time, many speech AIs have also been developed. People aren't just interested to have the computer play a descriptive game with them, they would like it to be able to fully converse. That would be way more of a big deal for people.
Already too smart to help anyone with their dead father's fortune.
For many years, the research and slow evolution of "conversational agents" continued (though the term "chatterbot" has gained greater usage among the general public).
In 1990, inventor Hugh Loebner decided to start a yearly contest for scientists to pit their programs against human judges. Since then, this has become like the Olympics or Academy Awards of the Artificial Intelligence field. The scientists who enter their creations are no mere geeks, but some sort of cognitively advanced supergeeks, able to trick machines into thinking they are human so the machines can trick humans into agreeing.
Not only are machines put to the Turing Test, but human subjects are mixed in to keep the judges on their toes. Yes, yammering humans are mistaken for bots frequently. That may not say much about the state of technology, but it probably suggests that some humans need to work on their conversational skills. A lot. When you are telling someone an amusing anecdote (in your mind), and they react as if they can't get their bank's computer voice on the phone to transfer them to a human already...you might want to think about it a little. Also, check with the bank, because they might be hiring.