Sep 25, 2014
At Automated Insights, we say that Wordsmith, our natural language generation platform, can write like a human expert.
And yet Automated Insights is employing me, a certified human, to write this blog post.
If we really make software that writes like a human, why can’t it write about itself? Can’t Wordsmith do its own PR?
Not yet. Let me explain why Wordsmith still needs me.
Wordsmith’s output is indeed indistinguishable from human writing. However, its process is decidedly non-human. The platform ingests structured data and then uses algorithms to find insights and express them in natural language. All of this occurs at a speed and scale no human brain could match.
But what is structured data? In simple terms, this is the kind of data that can be neatly represented in a spreadsheet. And while you can theoretically write anything you want in a spreadsheet, “structured” implies standardization that allows algorithms to parse the data.
Sep 18, 2014
Two of our data scientists, Adrian Atkinson and Ben Bruning, put together this analysis of what your fantasy football record means for your chances of making the playoffs.
“Everyone has a plan until they get punched in the mouth.” –Mike Tyson
Fresh out of the draft room, your fantasy hopes are usually riding high.
You’ve assembled a team with the star power, depth, and versatility to make a deep postseason run. All your bases are covered; you’ve got contingencies on top of your contingencies.
But, as we know all too well, almost everyone will have some rough patches throughout a long fantasy campaign.
It’s two weeks into the latest NFL season. With any luck, you’re the one doing the mouth-punching so far (to paraphrase warrior-poet Tyson).
If you’re currently sitting at a cool 2-0, congratulations—your plan has been executed to perfection. Based on 2013 Yahoo! data, the odds of your team extending that unbeaten start are as follows:
Chances of starting the fantasy season:
3-0: 1 in 7.6
4-0: 1 in 13
5-0: 1 in 23
6-0: 1 in 40
7-0: 1 in 71
8-0: 1 in 111
9-0: 1 in 203
10-0: 1 in 302
11-0: 1 in 441
12-0: 1 in 691
But, on the other hand, if your season’s off to a slow start, let’s crunch some numbers to determine how dire that 0-2 record really is.
Table 1 shows the empirical playoff probabilities for the 2013 Yahoo! fantasy football season.
That is, it illustrates the percentage of teams that ended up making the playoffs after starting the season with a given record through Week 1 or Week 2.
Unlike the NFL (where 12 out of 32 teams, or 37.5%, make the playoffs every season), your fantasy playoffs odds can be heavily influenced by league-specific settings.
Sep 15, 2014
The results are in: the Automated Insights Wordsmith platform really does write like a human.
This blog post presents the evidence in one place.
We’ve blogged before about Enter the Robot Journalist, a 2014 study in Journalism Practice by Christer Clerwall.
The study found Automated Insights content “objective” and “not necessarily discernible from content written by journalists.”
Furthermore, the study noted:
Although the differences are small, the software-generated content can be said to score higher on descriptors typically pertaining to the notion for credibility.
Google Search Results
I explained in a previous blog post that Wordsmith creates the kind of content that Google prizes: original, high value, and written for humans.
If that’s true, Wordsmith content should help brands perform well in Google search results – which, in fact, it does.
Our client Datatrac uses our Wordsmith content for SEO purposes. As of this writing, Google searches for “auto loans Raleigh”, “auto loans Richmond” and “auto loans Washington DC” showed Datatrac pages in the top five results.
If anybody knows quality content, it’s the humans who are paid to create it.
Matt McFarland, writing in The Washington Post about our fantasy football stories, says it is “not especially obvious that they were generated by a computer.”
Mashable’s Lance Ulanoff says Wordsmith produces “stories that can be indistinguishable from ones written by humans.”
And Slate’s Will Oremus writes this about Wordsmith’s tone:
You might think that what separates human writing from robo-journalism is the ability to write with flair. In fact, Automated Insights’ machines have little trouble couching their reports in a snarky tone if that’s what the client requests.
See For Yourself
Discover just how human our automated content can be with these two quizzes:
Quiz: Can You Tell Human Writing from Robot Writing? (Yahoo)
Robot vs. man — can you tell the difference? (Washington Post)
The Washington Post quiz compares Wordsmith with human writers for The Associated Press. The Yahoo quiz includes some Wordsmith content in its “robot” examples.
While it’s not a scientific survey, the majority of respondents to the Yahoo quiz scored below 60%.
The line between human and automated content is indeed blurry.
You may have already read Wordsmith’s automated output without realizing it.
For example, the automated corporate earnings stories Wordsmith generates for The Associated Press have the same style, tone, and accuracy as the human-written stories they’ve replaced.
The difference, as AP notes, is that the company can now produce 4,440 such stories each quarter rather than 300.
In cases like this, not noticing Wordsmith is one of the biggest compliments you can give it.
Contact us to learn more about how Wordsmith can automatically create human-sounding content for you and your business.
Sep 9, 2014
As Manager of Media & Public Relations, it is my pleasure to unveil the latest high-tech innovation from Automated Insights: our new Frequently Asked Questions page!
Click on over to learn more about our company and our natural language generation platform, Wordsmith.
You’ll get answers to questions like:
1. Why use natural language to present insights?
2. Can Wordsmith match my tone, voice, and style?
3. Where can I see Wordsmith in action?
4. Do we exist in one of an infinite number of parallel universes?
Please note that if you don’t see the question about parallel universes on the FAQ page, you must exist in a universe in which my boss made me take that one out.
If you’ve got questions of the less-than-frequently-asked variety, you can always contact us via email or on Twitter.
Sep 4, 2014
Our friends at Yahoo have just published some must-see fantasy football analysis by our data scientist Adrian Atkinson.
Adrian tackles ones of the most important questions in sports: how much smack can I talk if my team is projected to win?
Using NFL and Yahoo Fantasy data, Adrian masterfully illustrates just how confident you should be given your spread.
I won’t say any more here, but I will tease you with this graph:
Now head over to Yahoo and read the full analysis.
Sep 3, 2014
The rise of Big Data has been rapid and recent. Below are five Google Trends graphs that help tell the story.
First, a quick refresher from Google on how Trends graphs work:
Numbers represent search interest relative to the highest point on the chart. If at most 10% of searches for the given region and time frame were for “pizza,” we’d consider this 100. This doesn’t convey absolute search volume.
Our first two graphs indicate that the demand-driven data deluge shows no signs of diminishing:
The ascent of both “more data” and “too much data” matches the reality we explore in our recent white paper: despite massive data growth, there is a relative dearth of data analysis.
In other words, people have more data than they know what to do with. And that explains the rising popularity of our next two terms:
In recent years, “data analytics” has become slightly more popular than “data science,” although the relative popularity of both terms has grown significantly.
Perhaps driven by this growing awareness of data science and analytics, interest in Big Data tools and software has been exploding since early 2012…
…as has interest in data scientist jobs and salaries:
For more on Big Data, download our free white paper and learn how natural language generation turns Big Data into actionable insights.
Aug 29, 2014
It’s fantasy football season!
Our Wordsmith platform is gearing up to once again produce millions of personalized (and snarky!) Yahoo fantasy matchup reports.
Right now, Wordsmith is producing fantasy draft recaps and grades for every Yahoo team.
People are tweeting about their draft grade. Including these guys:
So what’s the deal with those grades?
In the video above, Joe Procopio, our VP of Product Engineering, explains what goes into a Yahoo Fantasy Football draft grade – and whether your fantasy draft grade is truly your fantasy destiny.
If you’d rather read than watch a video, here’s Joe’s wisdom in lightly-edited text form:
Aug 27, 2014
We’ve just published a new white paper, “Big Data Needs Big Insights: The Business Case for Natural Language Generation.”
As businesses spend big on Big Data, there’s a growing need to derive actionable insights.
Our paper explores why Natural Language Generation (NLG) platforms succeed where dashboards and data scientists fail.
"Big Data Needs Big Insights" explains how NLG platforms can deliver real-time, engaging, actionable insights for every level of your business.
Download our white paper for free right now.
Still not convinced you’ve got to read this white paper?
Here’s an excerpt to get you started:
Aug 26, 2014
Here’s the latest installment of our “Ask Ai” video series!
In this episode, I test whether our content sounds robotic. Can you tell which piece is automated? Watch and learn!
Got a question? Find and follow us on Twitter @AInsights.
Watch our previous episode to learn more about the “robots” in our office.
If you’ve watched the video and you want to read the content in question, see below.
Aug 20, 2014
Automated Insights CEO Robbie Allen takes the #IceBucketChallenge and will donate to fight ALS.
Robbie was challenged by Ai team member Brian Sewell.
Robbie is challenging three members of the Triangle startup community: Chris Heivly of The Startup Factory, Jason Caplain of Bull City Venture Partners, and Scot Wingo of Channel Advisor.