Robonews

Fw:Thinking
140 followers
13 Video Views·Dec 5, 2022
Good evening Mr. and Mrs. Robot from border to border and coast to coast and all the ships at sea.
And now, the news.
If you're like me, at some point in the day you're taking a look at news headlines.
It might be on Twitter, or on a website, or on a television broadcast,
or if you're really old school in this thing people call a newspaper.
You're looking to find out what's going on in the world.
You want it in a concise, understandable, and informative package.
Only a human could provide that, right?
Wrong.
Right now, as in this very second, there are artificially intelligent programs
writing news articles.
One such program is Heliograph, which makes the Washington Post its home.
Heliograph can take raw data and construct an article around it.
For example, in the United States in 2016,
Heliograph generated articles about political campaigns and elections.
So what's going on here?
First, the articles are more impressive than you might imagine.
They have an editorial voice,
meaning it's not just a string of cold, fact-filled sentences like,
Senator Bibbidi-Boo won 73% of all the votes in the district.
You might get something more like,
Senator Bibbidi-Boo held a dominant lead over all the other candidates,
earning more than 70% of the total number of votes.
Second, Heliograph isn't exactly donning a journalism hat
and writing these things all by itself.
It actually requires a lot of work on the back end
to allow the program to generate news articles.
Actual human being journalists must write up phrases
that the program can associate with certain outcomes.
They define the terms that appear in Heliograph's lexicon.
For example, I might create a rule that says
any candidate winning with 80% of the votes or more
is winning by a landslide.
Heliograph's approach is a fancy version of an old bit of logic.
It's the if-then statement.
If condition X is present, then outcome Y happens.
So Heliograph isn't thinking like a journalist.
It's essentially generating sentences
based on thousands of if-then rules.
So why do it?
It's like creating a form letter.
Sure, you could write a fresh letter
for every time you need to invite people to an event
or alert a job candidate that they didn't get a position,
or you could create a more generic form letter
that allows you to fill in fields
such as the recipient's name and save a ton of time.
That's what Heliograph does.
It saves time.
Instead of giving talented journalists
mind-numbing assignments, Heliograph can step in.
That frees up the human journalists to investigate
and write about more complex stories.
They can get to the deeper truths
and share them with the world.
That's what journalism is supposed to be about.
So Heliograph isn't about kicking journalists out of a job.
It couldn't do that because it's not flexible enough.
Instead, it can handle the numerous smaller stories
that develop rapidly over time.
And because it can generate these articles almost instantly
after receiving the relevant data,
it can help The Washington Post stay current,
even in a world with a 24-hour news cycle.
Will we develop AI programs
that will one day replace journalists entirely?
It's possible.
It will require an enormous amount of work,
and I suspect neural networks and deep learning
will play an important part in the process.
But until we can develop a machine
capable of simulating human insight,
I'm pretty sure reporters are safe from the robot revolution.
At least for now.
What do you guys think?
Which creative occupations are the ones
that robots and programs will master first?
I want to hear your thoughts in the comments below.
Thanks to Toyota for sponsoring this episode.
If you enjoyed it, hit the like button down there
and subscribe to our channel
to join the forward-thinking think tank.
Can I take my break now, Mr. Robot?
No?
Okay.
