Artificial Intelligence

-Mark Prus, Principal, NameFlash Name Development

About 6 months ago I wrote a blog post about the future of name development and the use of Artificial Intelligence to name things. I also made a prediction that AI was going to get better and better as it practices name development.

Since that time, AI is getting better at developing names. Check out these potential names for tomato varieties from a recent post by Janelle Shane.

  • Floranta
  • Sweet Lightning
  • Speckled Boy
  • Flavelle
  • Pinkery Plum
  • Market Days
  • Fancy Bell
  • Mountain Gem
  • Garden Sunrise
  • Honey Basket
  • Cold Brandy
  • Sun Heart
  • Flaminga
  • Sunberry
  • Special Baby
  • Golden Pow

I’m pretty impressed by some of these!

Of course, AI also generates names that might be considered to be bad choices, like these examples:

  • Birdabee
  • Sandwoot
  • Shampy
  • Bear Plum
  • The Bango
  • Grannywine
  • Sun Burger
  • Bungersine
  • First No.4
  • Smoll Pineapple
  • The Ball
  • Golden Cherry Striped Rock
  • Eggs
  • Old German Baby
  • Frankster Black
  • Bumbertime
  • Ranny Blue Ribber
  • Adoly Pepp Of The Wonder
  • Cherry, End Students
  • Small Of The Elf
  • Champ German Ponder
  • Pearly Pemper
  • Green Zebra Pleaser
  • Flute First

As predicted, AI is getting better at developing names. And this should increase the demand for professional branding services by experienced human beings! When AI was generating bad names it was easy to separate the stupid names from the barely acceptable names.

But now that AI is doing a better job, clients will have a harder time choosing a name because the list of 100 names contains 85 good ones! Clients will need the assistance of a branding expert to help make a branding decision, and perhaps a research expert to get consumer feedback on names.

-Mark Prus, Principal, NameFlash Name Development

Janelle Shane is a research scientist who likes to play around with neural networks. Recently she’s been having fun investigating whether neural networks can replace traditional means of creative development. As a professional name developer, I’m watching her work closely because I’ve been told that my chosen career is about to be destroyed by the use of artificial intelligence to develop brand names.

Based on the results thus far, I’m not worried. While it is true that computers can develop names, I strongly believe that the judgment of a seasoned branding expert (like me!) will be necessary to identify names that will resonate with consumers. As evidence of my confidence, I provide some examples of names developed by artificial intelligence in the past year:

  • Paint Colors – Janelle’s experiments yielded names like Stoomy Brown, Stanky Bean, and Bank Butt. I’m pretty sure nobody would buy a paint called Stoomy Brown (which actually looks like a shade of green) or Stanky Bean.
  • Craft Beers – The AI developed names like Toe Deal, Sacky Rover, and Cherry Trout Stout. Given the proliferation of crazy craft beer names, some of the names developed by the neural network appear to be reasonable (e.g., Devil’s Chard, Whata Stout, and Black Morning), but you have to sift through a lot of “Toe Deals” before you get to a decent name.
  • Guinea Pig Names – While the AI names for guinea pigs are better (e.g., Funbees, Sporky, Furzy, and Farter) that is only because you generally don’t have to say the name in public. Can you imagine using “Farter” as a dog’s name? “Stay Farter!”
  • Superheroes – I really don’t think a superhero called Nana will be feared by an evil villain…although I’m heard of some pretty badass grandmas. And would Supperman’s superpower be the ability to put the fear of bankruptcy in the hearts of owners of buffet dinner establishments?

You get the point. Right now it is all fun and games and it is easy to separate the stupid names from the barely acceptable names.

But eventually the AI will get better, and that is when demand for my services will actually increase! When AI starts generating excellent names companies will be faced with having to pick a name from a list of 100 great names, and they will need the assistance of a branding expert to make that decision. Put me in coach…I’m ready to play, today!

While on my flight back from speaking on trademarks & the alcohol industry at this year’s CiderCon, I listened to a fascinating TedTalks podcast on advancements in artificial intelligence that you can find here.  Some of the examples of the progress made in robots included the concept of open-ended, non-linear “generative” thinking by advanced AI machines.  As discussed in the podcast, artificial intelligence is now at a level of sophistication where, when given a problem, the robot can create what sounded like an inventive solution for new designs.  This seems to be beyond just more efficiently and expeditiously completing calculations, but actually derived from robots “thinking.” And that got me thinking – who should be the author or the inventor of that new design?

jetsons-rosey-mack

Under U.S. patent law, “whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor” as long as the invention is novel and non-obvious.  An “inventor” is defined by the statute as an “individual” with no explicit qualifier that the individual be a human.  The basis for patent law – Article I, Section 8 of the Constitution – also doesn’t specifically require a human inventor.

Assuming that the robot would be the “inventor”, would the new design be assigned automatically to the owner of the robot?  Maybe, as it would be similar to an employer-employee relationship under current law.  But what about any claims by the manufacturer of the robot?  Perhaps this issue of ownership in patentable inventions generated by these robots would need to be handled through user agreements for the robots.  Or should the inventor be the human directing the robot, as a tool, to provide a solution?

Under U.S. copyright law, the copyright in a work vests initially in the “author or authors” of the work.  The robot situation seems similar to “the monkey selfie” copyright infringement action.  There, a federal judge dismissed the case on the basis that the statutory language did not plainly state any rights for non-humans, leaving it up to Congress and the President to decide whether a monkey or another animal had standing to sue for copyright infringement as an owner.

The law is often slow to catch up to technological advancements.  How do you think the law should develop in this area?

 

 

A world without trademark attorneys… frightening, isn’t it? (maybe more for me than for someone who isn’t a trademark attorney). It seems unlikely, but we may be inching closer to this apocalyptic scenario.

In Australia, a group of researchers, legal academics, and other organizations are collaborating on the creation of a computer program that they hope will remove the subjectivity from trademark infringement analysis. The goal is to develop a program which can calculate a similarity score between two marks. The score could be relied upon by businesses in choosing names and courts in determining trademark disputes.

This certainly isn’t the first time a computer has taken on tasks normally reserved for humans. IBM famously created computers that could beat world chess champions as well as win $1 million on Jeopardy!. Computers have also been able to discover laws of physics that took humans centuries to unearth. Oh, and don’t forget the more recent creation of a self-aware Mario.

But for all the trademark attorneys out there, please, take a deep breath. We’re not unnecessary (yet). In fact, most trademark attorneys would welcome the opportunity to remove some of the subjectivity in trademark analysis in order to be able to provide more certain advice to their clients. Unfortunately, removing all of the subjectivity is likely impossible.

Trademark infringement involves more than just the similarity of the marks themselves. The legal analysis involves balancing numerous factors. The type of goods or services sold is a major factor, and others include the channels of trade in which the products are sold, the sophistication of the consumers, the number of similar marks used on related goods or services, the intent of the defendant, and others. Thus far, the Australian project appears to be focused solely on the similarity of the marks.

Also, the scope of protection granted to a trademark can expand or contrast over time. The protection can even disappear entirely. We’ve lost a lot of good brands over the years to genericide, including every day words like aspirin, thermos, escalator, dry ice, cellophane and others. Even when trademark rights aren’t lost in their entirety, if similar marks are in use by third-parties without evidence of confusion, then the rights afforded to one owner may be more narrow.

If the project is successful, the computer program could serve as an additional resource for trademark attorneys. It could be a helpful tool in providing some objective assessment of risk that marks might be considered similar. If the “similarity score” is a 3 out of 100, it is likely a good choice. The score of 99 will hopefully be enough to deter your client from investing in its idea for Starbux coffee shops. But will businesses, lawyers, or courts really feel comfortable relying on a score of 60? Or 40? The program might help confirm the clear cases, but is unlikely to be helpful in the close calls, where objectivity is needed most.

On the other hand, the program could be helpful where the other factors weigh in favor of one party. A high (or low) similarity score would present another piece of objective, factual evidence that could help in motions for summary judgment. It would also be a lot cheaper than hiring a linguistics expert to analyze the marks.

The software could also be utilized in-house or by marketing and advertising firms to analyze new name candidates prior to providing it to the decision makers. Providing the similarity score for those high risk names could help avoid emotional investment in a name that is all-but-assured to be turned down by legal. I don’t care what your selling, the name “Nike” is going to be an uphill battle that you’re better off avoiding.

The project is likely to provide some very helpful insight in the future. But I think I’ll come into work tomorrow, just in case.

Jesse de Agustin, Strategist (follow Jesse on Twitter: @JdeAgustin)

Imagine: I am struck by an 18-wheeler on the New Jersey Turnpike and fall to my unfortunate death. While I’d be dead, pressing legal – and philosophical questions remain. Does “Jesse” continue to “persist” across time through his online profiles? Do they turn into any one else’s property after my death? While this inquiry might seem immaterial, when philosophical theories of personal identity are placed under the social media spotlight, this becomes a pressing issue.

I came across an excellent iTunes U, University of Hertfordshire podcast by Patrick Stokes exploring these issues. In his talk, “Do the Dead Live on in Facebook?” Stokes argues that while people are ‘. . . anchored to their bodies, it doesn’t mean their online identities are not extended dimensions of our offline identity. . .’ Therefore, Stokes argues in a way, persons “survive” their death through social media profiles even though ‘selves’ do not.

Facebook for instance, allows the deceased users’ page to be memorialized. Confirmed friends may write comments on the wall but the deceased is not searchable or open to friend requests. Only immediate family can request the removal of the account. Yet this seems odd; that simply by virtue of being dead in the “physical” world the person must somehow live on via social media. In other words, why is “memorializing” Facebook’s default option? 

Yet if social media profiles are “extensions” of one’s personal identity in the digital space, it’s odd to say that a person could “live on” through their profile without the actual “self” actively experiencing that survival.

Imagine if one day, Artificial Intelligence could operate my accounts after death, in a similar fashion as I currently do. If this becomes the case, it seems that the “human” element has been stripped from social media if a non-human entity continues to “manage” my profiles.

One company, Intellitar allows persons to upload photos, information, and the Artificial Intelligence system allows one to “interact” with someone who is dead, to a certain degree. This is fascinating yet rather creepy; and beyond individual profiles, this presents an interesting conundrum for brands. If social media effectively gives brands a “human” voice and provides a vessel into a brand’s “conscience” then it seems that a brand risks loosing this human personalization thanks to Artificial Intelligence – even though the machine would come across as a thinking, but not human entity.

What’s your view?