Catch up on stories from the past week (and beyond) at the Slashdot story archive


Forgot your password?
Check out the new SourceForge HTML5 internet speed test! No Flash necessary and runs on all devices. ×
It's funny.  Laugh.

Computers Make Good Ad Execs 41

philg sent us linkage into a really amusing little story running over at Yahoo about computers generating advertising. Its pretty amusing since it showed that patterns could be used to generate commercials that were just as creative as ones that came from actual creative people. I worked at an ad agency for a few years, so this one made me grin.
This discussion has been archived. No new comments can be posted.

Computers Make Good Ad Execs

Comments Filter:
  • I dunno, it sounds more like this proves that the enginneers who wrote the program are more creative than Marketing people. Which seems obvious to me, of course... :)

  • and we can live much happier lives unblemished by advertising meta-messages that we are not whole, good, happy, trendy, and hip if we don't buy this or that piece of schlocky crap made in some country that uses slave labor. Ads are designed to make you feel bad -- until you hand your money over.
  • give me ten of these things and let me make a beowulf cluster out of them. let me mess with them for a month. at the end of the month i will give you the complete works of hamlet, a copy of the windows 98 source code, three parodies of "the blair witch project" and several /. articles on the windows "NSAKey" thing. allowed that you generate as many random thigns as you want, and can throw away the ones that don't make sense, anything can look impressive
  • by Zoinks ( 20480 )
    This is no big deal. Program a computer to generate ads (or poems, or Irish jigs, or images) according to certain rules, and assuming you did a good job programming, the output will always adhere to the rules.

    The really big thing missing is that humans evaluated the output to decide what was ``good'' - not the computers. It's no different from feeding words into an anagram generator and only choosing the anagrams that you like.

    On the other hand, if they actually had something similar to a chess or backgammon playing program (generates possibilities and evaluates their merit), then we might be in trouble.

  • I find that the most memorable adds are typically of the humourous variety, but the article never mentions, as far as I recall, any mention of humour in the adds that the program creates. It's also interesting to note that the article seems to imply that the authors (of the program) found all the rules for creative advertisement creation (that may not be what they mean, but that's how it sounded to me).

    If they can come up with a program that could produce some humourous adds on demand (i.e. I want an add for product X and it has to be funny), I'd be very impressed. Of course, this is, as the AI crowd has long babbled about, much more difficult that coding up a few rules.

    As was posted earlier, it's much more impressive that these folks could figure out some of the "rules of creative advertisement creation" that the fact that they wrote some little program to prove the point.

  • If you want to see state-of-the-art artificial intelligence for yourself, check out Forum 2000 at

    I'm sure you will be amazed at what these computers can think up ;-)
  • I must admit the idea of Don Corleone doing a cat food ad is quite nifty!

    However, that's not creativity. It's just concept association. Ultimately, it's the human filtering the results that applies creativity: creativity is not just coming up with unusual ideas, it's understanding how these ideas play out into a coherent result.

    For instance, the ad about Dracula and cat food, well, sucked. The idea of Don Corleone is a bit cheezy, but it works. Taken together, these ideas don't show creativity. It's when someone recognises which is the good one and which is the bad one.

    So, we're back to square one: computers are not creative in themselves, they're assistants to creativity. In this case, the ad execs were so empty of creativity (is that surprising?) that they thought the computer did all the work.

    "There is no surer way to ruin a good discussion than to contaminate it with the facts."

  • Or, again drawing an analogy from evolution: we're only seeing the successful selections, so we gravitate toward the idea that there must be a method to the madness that produced these ideas.

    Think of all the ideas these programs spit out that didn't work, or, more likely, didn't really make sense. We don't hear about them, so we get the headline "computers as creative as ad agencies."

    Bah! I say. It's just like looking at humans and saying "look how adapted they are! There must have been intelligent design!" without remembering that history is littered with the fossils of other of nature's blindly conducted experiments that didn't work. [take that, mush-minded creationists!]

  • Consider the following. An "ad trend" is created by a fairly decent ad-exec. An "ad trend" here being an advertising style which is imitated by many. Now, at this point we have a good number of trends to choose from. A computer is programmed with a simple database of NOTs. For example, if looking for an ad about a product for aging women, I'm *probably* not going to create an ad with three frogs on lilypads saying "Men", "O", and "Pause". With a simple list of "NOT"s one could have a machine render full ad schemes, simply by inputting the type of product to be advertised.

    Is this creativity? Hell, no. The creative part was done by the human who firsted used style of advertising.

    Is this the Microsoft "Frontpage" of advertising?

    I hope not.

  • Computers Make Good News Program Execs:

    The same could be applied to the predictability of
    most network produced news programs. Instead of a product it would be a topic and instead of an image association it would be a spin factor.
  • One statment I found very... odd about thhe article was the following:

    Contrary to popular wisdom that creativity thrives on unlimited freedom and unconventional thinking, Mazursky and his colleagues discovered that rules were essential to producing the ads the judges found creative.

    Huh? Are there really people who believe that "creativity thrives on unlimited freedom"? Sorry, but it ain't so. I've always felt that the truly creative are those that are able to work within the rules, and still come up with something original and interesting.

    There are many art forms that impose strict rules and structure including many forms of poetry (sonnets, haiku, etc.), most forms of music, dance and architecture. Without some sort of structure these become babbling, noise, convulsions and heaps of rubble. Structure is good. The truly creative can work within structure to create beuty. The non-creative wannabe's whine about being repressed.
  • The article has it wrong, so here's the real one:

    Unfortunately it doesn't seem to be open to the public anymore. When I tried it yesterday it was still open.

    Basically, it's a database of images, and what sort of emotions/ideas they evoke. And how they are connected to each other.

    It was interesting, although I ended up creating an ad for hygenic pads that consisted of an elephant wearing a tennis skirt.
  • by FascDot Killed My Pr ( 24021 ) on Wednesday September 08, 1999 @03:17PM (#1694566)
    First, the examples given are a little misleading. What, exactly, does it mean for a computer to be "creative"? Clearly they didn't "program human psychology" into a desktop PC and have it spit back out "bullet-shaped cars suggest speed". More likely they just created a computerized version of a madlib. Get a bunch of nouns, verbs and adjectives. Pull at random. Evaluate the "idea" thus produced.

    Which brings me to my second comment: This type of "creativity" is exactly analogous to biological evolution. The ideas generated by the computer map to mutations in a genome. The humans doing the actual selecting (note the word!) of good from bad map to natural selection. I suspect the hit rate of the computer is probably on the same order of magnitude of the chances of a random mutation being beneficial.

    It is well-known that mutations role in evolution is not all that big--selection does all the real work while mutation just provides (some of) the raw material. Conclusions to draw are left as an exercise for the reader.
    Put Hemos through English 101!
    "An armed society is a polite society" -- Robert Heinlein
  • Computers will probably create within the bounds and rules given but it will not be innovate still. Breaking out of the box is good, but not too many suceed in doing so. I've done graphic design, ads and logos personally and I know how hard it is to try and be innovative to create something fresh and new.

    But these computer generated ideas may be great for brainstorm sessions as well since you might be brain dead one day and decide to ask your computer. If you don't like to admit using the idea it gave you, you can still build on it if nothing else.

    Computers, in the end, are still just tools and toys. But they are fun tools and toys. =)

  • Having read some of Edward De Bono's work, this doesn't suprise me. Advertising would be a place where lateral thinking should be valued highly.

    Lateral thinking can be as simple as concept juxtaposition, or in this case, replacement of a concept with something else representing it. This is a pretty simple technique, as are most lateral thinking techniques. They don't seem too hard to do on a computer given some effort.

    I think it's important to note that these are just ideas that come out of a lateral thinking session, and like all ideas, they then have to be filtered through a human's logical thinking processes, just the same as these computer-generated ideas do.

    Of course, it doesn't necessarily bode too well for ad brainstormers.
  • Not the story link, but that link to the ad. generator that it has. Even though the link is a bit wrong, it's obvious how to use it, and the site seems to be not responding.
  • Don't forget what happened to those who stayed behind.

    They all died in an epidemic that originated from an unsanitised phone.
  • This reminds me of an excerpt from Raymond Kurzweil's Age of Intelligent Machines [] (2nd book in a row for me that's out of stock at Amazon but Chapters has got it) where examples from poetry generated by a computer are compared to human poetry. Supposedly the s/w was good enough to fool a significant group of people.
  • creativity is suppose to be something which pushes the realms of currently accepted notions, not rehashs old concepts. This research simply shows that a computer can rehash old concepts within different scenarios, not that it can creatively produce something de novo. further to this comes the idea of appeal - creativity can be appealing or appauling ... dependent upon the audience ... ;)
  • by Paul Johnson ( 33553 ) on Wednesday September 08, 1999 @07:12PM (#1694574) Homepage
    For example, "Intel Inside" was heralded as this marketing marvel, yet I see it and think: "Ok. Great. Thanks."

    The reason that "Intel Inside" is a marketing marvel is not the cheesy logo and four electronic twonk noises, its the way that OEMs are pushed into including it.

    Intel provide a cashback deal (essentially large discounts) to OEMs who include the Intel Inside logo in their advertising. Part of this is that Intel get to review and approve the adverts before they go out. This lets them place the logo, but it also makes sure that the voiceover says "and its got a 400MHz Intel Celeron processor" as if that were a major selling point.

    Viewers think that they are being sold a computer, so they discount what they see about the computer. But the fact that the computer company seemed to think that a Celeron is a major selling point sticks in their heads, and makes them more likely to pick a Celeron-based PC no matter which brand they buy.

    It also lets Intel play fast and loose with the anti-monopoly laws. If you offer AMD processors as well, you might suddenly find all your ads being rejected by Intel, and having to pay the full list price for your Intel CPUs.


  • Researchers at the University of ______ recently pitted a Commodore 64 using a slightly modified version of the well-known Eliza program against an ad executive in the Turing Test, the canonical method of examining artificial intelligence. Details of the results have been withheld, but this reporter has learned that the aging 8-bit machine has since been granted citizenship, while the executive has been fitted to a 19-inch rack and configured to act as a firewall for a small business network.

  • I think all this shows is that human judges are confused as to the definition of "creative". If creative means "formulaic and based on what has already succeeded" then the computer wins. If creative really meant what it should, the ability to create something new, different, and effective, the computer would obviously lose. Pop culture has more or less destroyed the possibilty for any modern "creativity" to get the credit or recognition it deserves. No, I am not an artist or anything, in fact I feel bad for those with such talent, because of things like this.

  • I don't even the whole "substitution" method of making ads to begin with. Unless it's REALLY clever, they usually aren't that good.

    My favorite ads are the ones that either make you laugh a LOT, or make you say "That is really clever". Substitution generally does neither. Car--Bullet? Mosque--Tennis Ball? So what?

    My favorites are usually the super bowl ones, because they aren't so concerned with getting the point across that their product is "classy" or whatever as they are with making it stick in the viewer's mind. Let name recognition do the rest. I don't CARE what morphing effect you can do to make your car synonimous with a first class plane ticket or whatever. It won't make me buy the car. What just might make me buy the car is if the company has some very high name recognition (combined of course with a reputation for quality). Ads, no matter how clever, cannot create a reputation for quality, because they are coming from the company selling the product. But they can generate name recognition.
  • All this article says is that ads are so repetitive that they should be automated; leaving humans to do things that actually require an imagination (if such a thing still exists). It's scary that whole educational programs, such as the Ringling School of Art and Design's animation program, revolve around making ads. I say any product that can't sell on word of mouth alone isn't worth selling.
  • Contrary to the reports I've read on this, and quotes from the researchers, this experiment doesn't show that "computers make better ad execs than people" or that "creativity thrives on limitations rather than complete freedom".

    Rather, what they seem to have demonstrated is that a computer that is programmed with one method of creating advertisements can create better ads than some people who know absolutely nothing about writing ads.

    Gee. Are we also shocked to discover that Microsoft Word's spell checker can spell better than someone who is completely illiterate?

  • by darklink ( 79588 )
    you mean now marketing is a computer , i hope it isnt y2k complient.
    sorry but i dont enjoy adds as is , i would realy hate to see a computer generating commicals for tv , inet , news papers , that is all i need is junk mail from a computer capable of shooting out 1million a second any way , now if this was ai or any thing i would still be agianst it this use is not good for paper or electrons, why make something that every one hate no one wants , but every one pays for . i mean seriouly i cant seem to figure why any one would make such a thing so annoying as this.
  • actually, I think that was how the code to win98 was made... (and unfortunately, I've heard that they were thinking of cloning BillG to aqcire win00 We sure knew he was a zero, but a doublezero... Oh well, that's only logical I presume...)
  • And I also wonder why the article refers to authoritities (i.e the judges) as if they were something to rely on. This is called to argument ad verecundiam and has got nothing to do in an article. What matters in an article is the arguments that the writer come up with... I say; if I said something like this; "Win98 must be good, 'cause BillG says so."; would anyone take me seriosly? I really do not hope so...
  • ``We are still quite far away from a possible conclusion that humans can be overtaken by computers, mainly because computers can not criticize their own ideas,'' Mazursky said. ''However, our evidence suggests that computers can at least aid and support the creative process.''

    I'm sure, with a few more focus groups, anything over the IQ of a use tea bag could be more creative than mere "humans." A person is smart, people are not. This is just another affirmation of that :-)
  • Did anyone actually see the ads that the judges chose?

    Other computer-generated ideas included an image of a bullet-shaped car, suggesting the car's speed, and a cuckoo shaped like a jet plane popping out of a cuckoo clock to show an airline's punctuality.

    Are these judges on crack? A bullet-shaped car? A cuckoo clock? THESE are supposed to be creative? Those are some of the worst ideas I've ever heard. However, they're right on par with the mediocre I see from most ad agencies.

    For example, "Intel Inside" was heralded as this marketing marvel, yet I see it and think: "Ok. Great. Thanks."

    The issue isn't that the computers are that creative, is that the ad execs are not. I know. I've been in two many meetings where horrible ideas like this were well-received by managers.

    - Scott

    Scott Stevenson
  • hmmm... mix and match images, select only those
    matches emphasizing a shared quality, and have a
    human create the final product?

    there's at east a dozen such programs already out there. paramind does something like this, for
    example. I've gotten even better results just
    using emacs and dissociated press... generated
    some damned cool surreal poetry.

    the important step is the *filtering* process
    after the phrase generation.

    also used various random midi generators as part of my experimental music compositions: create
    midi streams and mix them together to "paint" an aural picture.

    pretty nifty. but nothing new.
  • The researchers found that 89 percent of award-winning ads match as few as six formulas, which they called "creativity templates"...

    Could it be that the problem isn't supergood computers, it's that human designed-advertising is uncreative drivel? Kinda sounds like it.


  • At one time i worked for a very horrible place that happens to make yearbooks for most of the elementary and middle schools in the U.S. (if you have any idea what i'm talking about it wasn't Josten's it's the other one) When I started it was an assembly line arrangement where an image aquisition team scanned and named all the pictures and a quark team filled in the blanks on a template.
    Eventually that was inefficient so a plotting team took over and code warrior scripts were used to build books, place images, do everything but tweak the thing into final submission
    When I left there was a project to get rid of the final human element which was designing the covers for these books. The script would match the school mascot randomly pick between fractal and other backgrounds, try and make a guess as to whether it was a religious school ie. if it saw st. anything it would drop some crosses and sacred hearts. They didn't have an astounding success rate and the amount of time you spent letting the maching do it then re-doing the job yourself 75% of the time made it unworthwhile but everything evolves..
    I don't think this is the death kneel of the ad exec but i think in the near future it's gonna clear the field quite a bit..
  • I look through a lot of comments and no one criticzed the way these people did their research. See, from reading the second hand article (and I do realize that they do not say everything and they embellish, etc.), they did not use a real scientific process and they do not understand the meaning of creative. Couldn't the real creative process going on there be coming up with the idea to merge the product with a picture of an image of what you want to portray. Not the actual picking of the image. If this is the case, then they were already putting the creativity into the machine via the creative templates. THus, nothing they said can even be taken seriously. Summary: It takes creativity to come up with a creative template, not to fill in the void of the template. Plus, who cares what the judges say, they might be poor judges of creativity. And I want to see how they did everything (hopefully they did not say what I wanted to know and I just read over it). Did they submit several hundred, to exaggerate, ideas from the computer and the creative ad people only submit a few, so by shear luck the computer came up with better, or just as good, ideas. I mean, can a creative person consistently come up with good ideas, whereas the computer can not, something the article did not talk about (I know they talked about the computer coming up with junk but not about the realtive numbers of how much junk the humans come up with).

"An entire fraternity of strapping Wall-Street-bound youth. Hell - this is going to be a blood bath!" -- Post Bros. Comics