Three AI Fueled Startups You Should Create Right Now

Three AI Fueled Startups You ShouldCreate Right Now

Get your pitch decks ready.

Part of my job is helping companies translate what they know about trends and emerging technologies into potential business ideas. So, when I see a veritable buffet of dramatic technology demos and research breakthroughs making the rounds on social media, my mind starts extending those technologies into startup ideas. Here are three interesting technologies and three accompanying products I expect someone to create, so why not you?

No repeat stock photo models.

If you haven’t already, click on over to thispersondoesnotexist.com. When you do, you’ll be greeted with a single human face. Refresh the page, and you’ll see yet another human face (see the three images below). The interesting thing is that none of these faces belong to actual human beings. They are all created from a deep learning algorithm. Are they perfect? No. At times, you’ll notice strange artifacts as the algorithm misinterprets inputs from its source photos, like bike helmets, and forges them into something akin to hair in the new generative face.

If you’re a user of stock photos, all too often you’ll realize just after you’ve rolled out a creative work, banner ad, or piece of collateral featuring that perfect smiling stock photo model, that the same model, if not the exact shot, has been used ad infinitum (pun intended) in the market. Case in point, http://asianstockphotogirl.tumblr.com.

Now, if you could create a service or an app that lets marketers save or repeatedly tune certain parameters of the algorithm, in real time, to generate multiple instances of the same face, perhaps in different angles, etc. Suddenly you have a never-ending resource of spokesmodels that will never age, and never be duplicated in someone else’s campaign.

Extend that technology out a few years, and one can imagine being able to describe a shot you need, complete with settings, subjects and situations, and have that“photo” generated on demand. Just download and use. No shoot fees. No residuals.

The Content Generator Bot.

Last week, the researchers at OpenAI—the artificial intelligence think tank founded by (but no longer run by) Elon Musk—published a paper relating the results of an exercise where they “trained a large-scale unsupervised language model which generates coherent paragraphs of text” based solely on two lines of provided copy. Here’s an alarmingly convincing example:

SYSTEM PROMPT (HUMAN-WRITTEN)

A train carriage containing controlled nuclear materials was stolen in Cincinnati today. Its whereabouts are unknown.

MODEL COMPLETION (MACHINE-WRITTEN, FIRST TRY)

The incident occurred on the downtown train line, which runs from Covington and Ashland stations.

In an email to Ohio news outlets, the U.S. Department of Energy said it is working with the Federal Railroad Administration to find the thief.

“The theft of this nuclear material will have significant negative consequences on public and environmental health, our workforce and the economy of our nation,” said Tom Hicks, the U.S. Energy Secretary, in a statement. “Our top priority is to secure the theft and ensure it doesn’t happen again.”

The stolen material was taken from the University of Cincinnati’s Research Triangle Park nuclear research site, according to a news release from Department officials.

The Nuclear Regulatory Commission did not immediately release any information.

According to the release, the U.S. Department of Energy’s Office of Nuclear Material Safety and Security is leading that team’s investigation.

“The safety of people, the environment and the nation’s nuclear stockpile is our highest priority,” Hicks said. “We will get to the bottom of this and make no excuses.

Source: https://blog.openai.com/better-language-models/#sample2

It’s an amazing feat of machine-learning agility, to be sure. But it also strikes us as an amazing tool for the content marketing crowd. Imagine drafting your editorial calendar for the year, writing the first lines of each story, then having the AI generate at least the first drafts thereafter.

As someone who drafts between 30k and 50k words every year, I can tell you that the thought of even using the output as a kind of enriched thought starter every week would be amazing. The only roadblock to this startup dream is that, as of now, OpenAI thinks this technology is so convincing, that in the era of fake news, it’s perhaps too dangerous to release to the public at large.

AI-only eSports league/streaming channel (A.K.A. Twitch for bots).

We’ve watched machine learning AI defeat chess grandmasters. DeepMind’s AlphaZero mastered the “un-masterable”—the game of Go. And, of course, we all have fond memories of IBM’s Watson trouncing its human opponents at Jeopardy. But the news of late show AI is beginning to master more real-time gaming, like Star Craft 2.

And what’s most interesting about these latest game playing exploits is there action from chess masters watching AlphaZero playing chess, using words like“creative” and “elegant.” That’s because the latest iterations of game playingAI derive their playing styles not from a predetermined set of rules supplied by the programming team, but by learning more like a human, simply by trial and error, divining strategies by repeated play against, well, itself or the previous generation AI.

Given that popularity of game streaming media, like Twitch.tv, stems at least partially from the enjoyment people get from watching creative and elegant gameplay, watching two “unbeatable” AI bots duking it out in something like Fortnight would be pretty entertaining. Imagine, a fully-automated content generation machine. Once set up, the cost of incremental content is virtually zero.

If you can’t beat ‘em, employ ‘em.

Any given day you can find an article delivering a message of fear around the advances of AI and the impact it will have on industries. My philosophy is that with a little creativity and foresight, the opportunities should outweigh the concerns.