AI fake news generators like GPT-2 could be used to muddy coverage of the U.S. 2020 presidential election, according to a new report.
The study — featuring survey results from 233 publishing industry execs — was released by the Reuters Institute for the Study of Journalism and the University of Oxford.
Execs surveyed worried that in the hands of expert operators, AI-driven fake news generators could churn-out significant amounts of propaganda and lies about presidential candidates.
But Joshua Benton, director, Nieman Journalism Lab, is less disturbed.
Benton tested the full-blown version of GPT-2 – released last fall.
He found that while GPT-2 can create readable prose, it’s not very good at creating believable prose.
Bottom line: While GPT-2 can generate news stories that are grammatically correct, most of its work product would not past muster with a highly skilled editor, according to Benton.
Currently, access to a number of fake news generators — including GPT-2 — is readily available, free-of-charge.
The Reuters/Oxford study touches on a number of important trends currently shaping journalism.
In other AI-generated writing news:
*AI-Generated Writing: Top Ten Trends for 2020: For a quick study on the state of automated writing, check out: “Artificial Intelligence Generated Writing: Top Ten Trends for 2020.”
Authored by Joe Dysart, editor, RobotWritersAI.com,
the piece offers an update on the tech’s key drivers, including:
–Newsrooms’ embrace of AI-generated writing
–Many writers’ fear that AI spells job loss
–Publishers’ assurance that AI is no pink slip for writers
–How AI is personalizing writing
–How AI is auto-writing company reports
–How advertisers are auto-generating short promotional
copy with AI
–Academic publishing’s adoption of AI-generated writing
–Public relations and its need to automate with AI
–Journalism students’ need to acquire AI-generated
writing skills
–The ongoing scourge of AI fake news generators
*New Report: The Best AI Writing Assistant Software for 2020: Grammarly Business, ProWritingAid and Ginger are the best choices for AI writing assistant software, according to a report from G2 Crowd.
“Rankings on G2 reports are based on data provided to us by real users,” says Michael Fauscette, chief research officer, G2. “They represent the voice of the user and offer valuable insights to potential buyers everywhere.”
The software review site is a key go-to source for software reviews.
G2Crowd strives to maintain objectivity in its ratings by refusing to allow paid placement from any software makers.
*Persado Gets Nod as ‘Cool AI Vendor’: AI-generated marketing firm Persado has been recognized by market research firm Gartner as a ‘cool vendor in AI marketing.’
Persado earned the distinction by specializing in AI- generating small snippets of promotional copy – including subject heads for marketing emails.
Persado currently services 250+ clients, including JP Morgan Chase, Humana and Vodafone.
“We’re still at the tip of the iceberg of what AI and natural language processing technologies can do,” says Alex Vratskides, Persado’s CEO.
*Daily Deals Site Scores Big With AI-Generated Social Media Copy: Daily deals Web site Wowcher significantly reduced the cost of its social media budget by tapping Phrasee for AI-generated writing.
Essentially, AI service provider Phrasee reduced Wowcher’s cost to generate sales leads on Facebook by 31%.
The solution Wowcher leveraged: Using AI to auto-generate ad copy for Facebook – rather than relying on human writers.
Generally, Phrasee auto-generates short promotional text for advertisers using a formula of create, test — then send-the-best.
With social media, for example, Phrasee generates multiple copies of the same social media copy, then tests each variation on a small population of social media users to see which variation works best.
Subsequently, the most effective copy is broadcast across social media.
Phrasee got its start with AI-generated writing in email marketing, where it uses the same create, test, then send-the-best formula to generate catchy subject heads for marketing emails.
*The New York Times Sees Uptick in Clicks-to-Stories With AI Personalization: The U.S.’ paper-of-record boosted click-through rates to news stories on its home page by 55% with AI personalization, according to Nick Diakopoulos.
Diakopoulos is assistant professor, communication studies and computer science, at Northwestern University.
Diakopoulos says The Times pulled-off the spike in interactivity by personalizing and localizing specific sections its home page.
Essentially, the AI programming enables readers to be served a news story personalized to their interest – based on where they live and other personal data.
Diakopoulos offers an interesting take on how AI is transforming newsrooms in this Digiday article.
*AI-Generated Writing’s Big Promise: Fortune Magazine offers a deep dive into the rapidly evolving promise of AI-generated writing — also known as natural language processing in this piece.
During the past 18 months, “computer scientists have made huge strides in creating algorithms with unprecedented abilities at a variety of language tasks,” observes author Jeremy Kahn.
“What’s more, these new algorithms are making the leap from the lab and into real products at a breakneck pace—already changing the way tech’s biggest players, and many other businesses, operate,” he says.
*First Steps: Adding AI-Generated Writing to a News Outlet: AI-generated writing firm United Robots has launched a promising ongoing AI series.
Its goal is to detail best practices for embedding AI-generated writing at a news outlet.
During the past four years, the company has added AI-generated writing capability to numerous newsrooms.
One of United Robots’ key findings, according to the company blog: “Swedish publishers have found that robot-generated, hyper-local sports and property sales coverage increases the relevance they can deliver to subscribers — which in turn drives engagement and thereby retention.
“That level of granularity and volume of articles would not be possible to produce through reporter resources alone.”
*Google Pushing for Measured Regulation of AI: As the European Union finalizes proposals for the regulation of AI, Google is calling for restraint.
Sensible AI regulation needs to take a proportionate approach, “balancing potential harms with social opportunities,” according to Google CEO Sundar Pichai.
In the U.S., many technologists are pushing for a similar approach.
David Edelman, for example, is recommending the Trump Administration regulate AI implementations on a category-by-category basis.
That approach is far superior to attempting to govern with just a few broad strokes, he says.
Edelman is director, Project on Technology, the Economy & National Security, at Massachusetts Institute of Technology.
*A Go-To Source on AI and Journalism: With this four-minute video, Seth Lewis shares his penchant for examining how AI and similar technologies are changing journalism.
Lewis is an associate professor in emerging media at the University of Oregon.
*Just For Fun: For people who yearn to send a personalized greeting card — but just don’t have the time — there’s Handwrytten.
For a small fee, the company will take your typewritten well wishes and have a robot hand-write them for you on a greeting card.
Simply keypunch-in your heartfelt notions on the company’s Web site or in its handy app. A robot on the other will take the trouble to make them appear handwritten.
David Wachs, Handwrytten’s founder, says he had 100,000+ orders for his mushy missives during the last Christmas season alone.
COMMENTS: Have an opinion or comment on this article? Feel free to click the green flag to post your thoughts:
–Joe Dysart is editor of RobotWritersAI.com and a tech journalist with 20+ years experience. His work has appeared in 150+ publications, including The New York Times and the Financial Times of London.