Bull Session

AI Goes to the Ballpark

July 7, 2016          

Episode Summary

On The Digital Life this week we chat about technology and the great American past time, baseball.

Just last week the Associated Press announced that it’s covering Minor League Baseball games using AI software. The software from Automated Insights, draws upon supplied game data to create a written narrative. This AI is already being used by the Associated Press to create earnings stories on U.S. public companies and by corporate customers like Edmunds.com, which uses it to generate descriptions of cars for its Web site.

So, AI can cover a baseball game, parsing the data and creating a narrative, but is the writing any good? So far, it seems to generate stories that are readable, but not really compelling or interesting beyond the most mundane facts. Is this the future of sports journalism? Join us as we discuss AI and baseball.

AP Sports is Using “Robot” Reporters to Cover Minor League Baseball
AP expands Minor League Baseball coverage

Welcome to episode 163 of The Digital Life. A show about our insights on the future of design and technology. I’m your Jon Follett. With me is founder and co-host Dirk Nemeyer.

Hey Jon.

Hey Dirk. For our podcast this week, we’re going to talk a little bit about that great American past time, which is baseball. Dirk are you a baseball fan, or do you just watch the game on occasion?

When I was younger as a child, teenager, maybe young adult I was interested in baseball. It’s been a really long time now and I watch the game on no occasion.

The reason that we’re talking about baseball on a future technology show is that the Associated Press recently announced that it’s covering Minor League Baseball game, on a national basis, using artificial intelligence and software from a company called Automated Insights. This the software company takes the MLB data and the stats from Minor League Baseball and it basically generates a narrative that then goes over the AP Wire, and the AP even has an editor who’s specifically …

He’s or her job is specifically to look at this content that’s being generated by the AI, and making sure that it passes muster. They do have a human in the loop, but what’s really fascinating I think is this use of technology to create the human interface which is the text, is created by the AI. I kind of feel like, when you think about baseball in the long history of the game. You think about this all the human achievements that go along with it, and it’s also of course very important as part of our national culture.

To see this take place, which is essentially the AP does not have enough reporters to cover Minor League baseball games. Now it has this AI reporter, and if you don’t mind I’m going to read a sentence or two from a game that was written by an AI. Here it is.

Jon, do so in your best artificial intelligence voice.

Okay. I’ll try. God this is going to be bad. “State college PA, Dylan Tice was hit by a pitch with base is loaded with one out in the 11th inning giving the State College Spikes a 9 to 8 victory over the Brooklyn’s Cyclones on Wednesday.” What do you think?

I thought you sounded great, actually. Well done, well done.

You can see how the … It’s a very utilitarian first line there. It goes on to talk about the game winning run, and the double play that preceded it, etc and the score throughout the game. It’s very much a, it doesn’t even rise to a level of a junior reporter recording this because there is no flavor per se. At least there is no flavor injected yet into this reporting it’s pure data transferred to narrative. At some point in the future, I would hope that automated insights working conjunction with MLB will come up with some game flavor.

No human actually has to do what I did, and read that copy because it’s extremely dull. Now automated insights actually provides this service to other companies besides MLB including companies reporting earnings or if you have like edmunds.com I think uses them to describe their cars. They certainly got thousands and thousands of cars on their websites. They need some kind of narrative for each one of those. That’s generated by AI. One of the things that I enjoy about baseball is the narrative, the flavor of that, the tension, the writing frankly. For me, this is not compelling writing. Dirk what’s your take on this first 4A of transferring pure data into narrative?

I don’t have the right language from the standpoint of journalism and publishing. I’m going to … The way I’m going to communicate this, is going to be clumsy. There’s somebody who immediately is going to be other saying, “The right word for this is this.” The sort of short run down game summary structure. Even when it was written by humans have that similar sterile feel to it. Like there is in the context of sports reporting a genre is the wrong word. There are this is just a basic boring game.

Sort of here are the facts run down of it. Humans have done that crap for a long time, right? I don’t think it’s particularly remarkable or unremarkable, that what the AI is producing in this case is sterile. Because what the humans produced in this case is also sterile. My question is more, would the AI be capable of more. It’s very safe to have AI do the sterile thing. As you pointed out the article mentions that, and to me. This was the more interesting part of it, the article mention this has already being done for earnings report, and cars and other stuff.

None of that is getting big write ups, right? What is the sort of romantic about baseball and sports and our relationship with it. That now just becomes a story even though what’s being done here in the baseball context is just a per forma and blend and sort of lack of imagination and representation of just sort of human scale of cranking it on its scale. Yet now that it’s a baseball story, it’s a headline. it’s something that we’re able to be exposed to through the wonderful world of liking and re-tweeting.

To me that’s fascinating and particularly because you talked about the romance of the narrative and why that makes it important. What’s happening here has nothing absolutely nothing to do with that. Yet just the context of baseball and the tradition of romance and narrative, elevate this in a way that from a technological perspective from a theoretical job taking perspective the story was even more important at the earlier point when they started to do this for corporate earnings, instead of Minor League Baseball stuff.

It’s just kind of a lot of interesting things later in here to look at and talk about.

Yeah, yeah I agree with that. Just to be clear, this is Minor League Baseball. The AI is not taking anyone’s job here. They’re not going to send a reporter out. It is not the World Series. Although I would love, hopefully these guys at Automated Insights would give us some world series reporting as well, because I’d be fascinated to compare the two, right? You talked about the romance, or we’ve talked about the romance of the narrative. Part of that, at least for me it goes back to the idea of sports as the greatest on scripted drama, right?

We have the scripted drama which we love in television and the movies. What’s so attractive about sports event is that you don’t know how it’s going to turn out and nobody else does either. There’s this immediacy, this of human effort that makes us love sports.

I think even more than that. It’s sort of shared interest right? There’s this thing that many of us feel attached to, and are rooting for. It’s the classic RB being becoming part of something bigger than ourselves.

Right, and so to have some of that mechanized, it seems feels a little weird, but of all the sports, or the sort of larger leagues in the United States. Baseball, I think has definitely has a front relationship with technology, you have the statistics loving, sort of Bill James, Acolytes, Moneyball who sort of believe that the numbers can tell you quite a bit about, what a player can do and there are different ways to take advantage if you can understand statistics and sort of where the value in specific players maybe.

Now not all of baseball management thinks that way, there are few teams or there are probably very many teams at this point who are leveraging that technology. There’s always been this tension, where it’s kind of this unholy marriage, right? The statistics are not a replacement for the human eyeball so called. Baseball I think just because of its longevity, and because of its priority as the national past time so called, but now of course we all know that’s mostly football. Baseball sort of high positioned in our society, and has all contributed to that tension with technology.

I think for myself when I think of the reporter at the baseball game. I kind of harking back to like the 1950s, the guy in the fedora hat with the reporter’s notebook we all know that’s not happening anymore. Sort of in our minds I see him replaced with this little computer, or an iPhone but wearing the same damn hat. I don’t think so either. To follow on what you were talking about Dirk there’s this idea that computers take jobs, or that software eliminates job. There are little incursions, right into areas that can be highly creative like not this particular baseball story.

Baseball writing can be very creative, and we talked a few episodes back about AI going to art school and generating this full Rembrandt based on the masters, back catalog of lovely paintings and this AI was going to create this new painting. The computer, the software as creative force or as nominal creative this rising creative element in our society, as a creative person I have, I always find that a little dodgy. I know you may not have the same problems with it that I do, but I feel like this baseball story falls into that same area.

I think the creativity is the special human things, just it’s a myth. I mean creativity is just sort of the manifestation of something unexpected, something less straight forward. That’s accomplished by some combination, of need that just forces spontaneous innovation, or just by people’s processor’s people’s mind and problem solving machines. Our internal machines operating differently or unless unnecessarily better where the creativity comes from people who are operating differently either due to the wiring and the piping that we have, and or due to the way that they’re solving problems or the contact store, or a lot of different factors.

None of that to me is special or unique and it’s just a matter of time before the most creative pursuits that humans are able to express will be match and exceeded by machines. There’s a lot I don’t know on the engineering side, so I hesitate to put specific dates on it. I know for a fact, like there are somethings creatively, I’m really good at. I know exactly how I could communicate to an engineer, what the process is, what’s going on that gets me from zero to really cool funky, unexpected solution.

If I can do that, other people can do that. Once we’re able to translate that into commands for the machine. It’s game sudden match, and I don’t mean that from like a scary, “Oh my god are irrelevant perspective. Even though that’s one possible long term outcome.” Just from the perspective that this creativity isn’t special, it’s not protected, it’s not safe. It might be further out before the machines are able to get there, compared to the very sterile little minor league baseball game that you just related. It is going to inevitably come and so rather than feared out or chase against it we just have to think about what’s next for us.

Yeah, I do think there are certain things of course that, the human mind is quite good at, and certain thing which the AI can sort of easily take on and clearly transferring baseball data into narrative is one of those things that the AI can do. How they’ll handle football or basketball, or hockey, or golf or whatever the next thing is for AI sports reporting. I think there are aspects of human thinking that are not so easy for the software to replicate just yet. If you were planning on being a Minor League Sports reporter to get your telling the door, then you might have to look elsewhere because the APs got that filled with this piece of software now.

Yeah, I actually expect it won’t be that long, so I mean one recent major sporting event was the NBA finals. It was significant because there were this huge narrative works in the NBA this season around Golden State with all the wins, Lebron James trying to win it for Cleveland. Golden State up 3 to 1 in the series and then Cleveland comes back to win it, which is unprecedented in that particular sport. I think we’re probably close to a machine being able to tell the narrative of that story in a really good, and compelling way, that if the way you consumed that story was clicking on a random garden variety new site, CNN, Yahoo, Fox, ESPN, looking at sort of the headline story. Then it would be quite good actually.

I even suspect that they might be able to do that now. If not now, probably not now, very soon I think that’s possible. What I suspect will take longer, will be for that story to reach the level of the very best human sports writers. I don’t know who those people would be now. If you look, if you’re talking about like going back to the fedora hat days, somebody like Grantland Rice. I think if Grantland Rice was taking on that, and the supercomputer was taking on that.

I think Grantland Rice would have the better story. He’d have the one that was just, it was a little richer, it was a little more bring a little bit more to the table, little more humanity, whatever you want to have. I think from the standpoint of the mass market, of just getting a story that’s in a really nice narrative, emotional even job. I think machines are going to get there pretty fast. At the peak of human performance, it’s just going to take a longer time probably similar to the progress of games like chess, and go frankly.

Listeners remember that while you’re listening to the show, you can follow along with the things that we’re mentioning here in real time. Just head over to thedigitalife.com, that’s just one “L” in The Digital Life, and go to the page for this episode. We’ve included links to pretty much everything mentioned by everybody. It’s a rich information resource to take advantage while you’re listening, or afterward if you’re trying to remember something that you liked. You can find The Digital Life on iTunes, Soundcloud, Sticher Player FM, and Google Play. If you want to follow us outside of the show, you can follow me on Twitter, at @JonFollett. That’s J-O-N F-O-L-L-E-T-T. Of course, the whole show is brought to you by Involution Studios. Which you can check at goinvo.com. That’s G-O-I-N-V-O.com. Dirk.

You can follow me on Twitter at @DKnemeyer. That’s @D-K-N-E-M-E-Y-E-R. Thanks so much for listening.

That’s it for episode 163, of The Digital Life. For Dirk Newmeyer. I’m Jon Follett, and we’ll see you next time.

No Comments

Leave a Comment

Your email address will not be published. Required fields are marked *