I had a standard conversation yesterday with Eliza, a famous chatterbox.

Eliza: "So how are you doing today?'' Ahmed: ''I'm in top form.'' Eliza: "Do you believe it is normal to be in top form?'' Ahmed: "Why should I come to an idiot like you?'' Eliza: "Oh, I?'' Ahmed: "Yes you.'' Eliza: "What difference does it make if I?'' Ahmed: "A lot of difference because I thought you were some sort of psychoanalyst.'' Eliza: "Oh, I was some sort of psychoanalyst?'' Ahmed: "'Stop repeating what I say.'' Eliza: "Have you asked such questions before?'' Ahmed: "Reasonably often.'' Eliza: "I'm not sure I know what you're talking about.'' Ahmed: "Are you confused?'' Eliza: ''Would you prefer if I were not confused?'' Ahmed: "Quit.'' Eliza: "Nice talking to you.'' As you can see Eliza, a software program, has some conversation limitations which would soon clue you into the fact you weren't talking to a human.

Eliza was created in the 1960s by MIT professor Joseph Weizenbaum as an early attempt at simulating artificial intelligence (AI) in conversation. He described Eliza as a sort of "Rogerian psychotherapist'' that rephrased patient questions into questions of her own to replicate a conversation with a patient. Eliza was studied by sociologist Sherry Turkle who noted that people became protective of their relationship with her.

Eliza, a neat, small piece of software containing about 240 lines of programming, has sparked many offspring, known as conversation simulators or chat bots. The chat bots attempt to live up to the standard of the Turing test, named after Alan Turing, the famous British mathematician who was key in conceptualising and designing the first computers during World War II.

Mr. Turing's goal was the creation of AI. He believed that machines could be made to think. In a 1950 article entitled Computer Machinery and Intelligence appearing in the journal Mind he came up with the simple test that if the responses from the computer were indistinguishable from that of a real person, the computer could be thought of as thinking.

Since 1990 The Cambridge Center for Behavioral studies has run a competition around the Turing Test, ready to award a prize of $100,000 for the first computer that outputs responses indistinguishable from a human's. While no one has taken the full kitty, a prize of $2,000 is awarded each year to the computer judged most human-like among those put to the test.

For many years commentators have predicted that bots would take over the Internet as a means of disseminating and controlling the vast amount of information out there. But so far the bots have not taken over on the Internet, although they have become annoyances. Robots haven't lived up to the promise, or threat, of taking over our lives either although Sony's programmable robot dog was certainly a hit during the Christmas season for those who could afford one of the $2,500-plus toys. Japan-based Takara Co. has now joined in the virtual pet craze with the robot fish, selling for about $140.

Chatterboxes have also evolved into what Bermuda resident James Martin calls "virtual idols''. One such virtual idol in Japan is Kyoka Dale who has a cult following and was rated the number two pop star in the country at one time.

Recently I saw a programme on CNBC which featured an interview with T-Babe, a new virtual pop star created by Glasgow Records. T-Babe is being designed to appeal to young men who are fans of the Spice Girls. Sort of a clone of clones. In the interview, done with the interviewer talking to the computer model on a large screen, T-Babe said she was a fan of Madonna.

Along with chatterbots on the Internet you can also find softbots, searchbots, userbots, taskbots, knowbots, and mailbots (to answer your mail). The more horrible kind are the clonebots and floodbots that can shut down Internet servers. Other people have developed annoybots and hackbots.

Commercial sites are attempting to develop smart agents to make a type of assistant that has human attributes and will guide you through the buying process without actually having to deal with a person. These bots are more and more reacting to instructions that appear human-like even though the underlying computer programs cannot be classed as AI.

The advantages are obvious. Instead of hiring costly human help, simply stick a bot on the site. Chat rooms on the Internet, where people communicate with strangers, are infested with bots. It's ironic that perhaps people are unwittingly talking to a computer! Julia was one such chatterbox designed to flirt outrageously with men on the Internet chat boards.

You can download the original Eliza and communicate with other much more intelligent chatterbots at The Simon Laven Page at http://www.student.topli- nks.com/hp/sjlaven/classic.htm. I like Jason Hutchens' page at http://ciips.ee.uwa.edu.au/hutch/h al where you can talk or view conversations with his creation MegaHAL, presumably named after Hal, the computer that attempts to take over the spacecraft in Arthur C. Clarke's 2001: A Space Odyssey. MegaHAL is pretty good and funny but certainly not as intelligent, or dangerous as Hal.

Mr. Hutchens invites you to conduct your own Turing Test by letting your mum or friends have a go by telling them that you're talking to someone famous.

Other sites for information include http://www.computingcentral.com, http://www.bottechnology.com and http://www.botspot.com.

In the end the search for artificial intelligence has come down not to inventing machines that can be like humans but to creating software packages that can do tasks humans can't do very well, tasks requiring speed, persistence, time and that involved vast amounts of data. It's really about making machines smarter through what commentators call alien intelligence.

"The Net-Powered Generation: A Technographics Summit About Young Consumers'' conference is being hosted by Forrester Research April 26 in New York. For more information go to http://www.forrester.com.

Tech Tattle deals with topics relating to technology. Contact Ahmed at ahmedelamin yhotmail.com or (01133) 467012599.