A Close-Up on How Long-Form Automated Writing Works
Editors and writers curious about AI’s ability to generate long-form writing will want to check-out this piece by SEPGRA, an economic think tank.
The group decided to give GPT-3 — one of the world’s most powerful AI text generators — a run for its money by inputting one, simple phrase and asking GPT-3 to respond.
The phrase: “Write an essay about text written by AI.”
The resulting 900-word essay published in this article is emblematic of the tech’s current prowess.
Essentially: The piece begins with an excellent focus on the specific topic, but becomes ever-more generalized as the article unfolds.
In fact, by the close of the essay, GPT-3 completely veers-off into a discussion of AI’s oft-reported ability to beat the world’s greatest chess masters.
That’s an interesting fact about AI, to be sure.
But it’s also extremely distant from what was supposed to be the article’s focus: AI-generated writing.
The experiment points-up a hard fact many GPT-3 users have long known: The tech has the remarkable ability to reel-out sentence after sentence with robotic efficiency.
And its ability to generate shorts spurts of text — such as ad slogans, email subject heads, short poetry and the like — can often be spot-on.
But users still need to find a way to tame GPT-3 to ensure it stays on topic when it comes to long-form writing.
One final observation: While GPT-3 lost focus on the topic as the article unfolded, its writing style remained steady — and extremely polished — through-out.
The take-away: Once GPT-3 is engineered to stay on topic for use in commercial applications, it’s long-form, automated writing will easily compete head-to-head with professional writers.
(For an in-depth look at GPT-3, check out: “GPT-3 and AI Writing: Stunning, if Imperfect,” by Joe Dysart.)
In other AI-generated writing news:
*In-Depth Guide: Outranking: Beng Keat offers an extremely in-depth guide to Outranking in this piece.
Outranking is AI writing software focused on auto-generating copy that scores high in the search engines.
Observes Keat: “Organic search is the holy grail of any content marketer’s heart.
“The key to higher rankings is more research, better writing, and more optimization.”
Outranking delivers on that goal, according to Keat.
Interesting features of the software include:
~The ability to pinpoint the exact type of content you need on your Web site or other digital property to bring in your desired audience
~The ability to know how Web sites competing with yours are faring with their content
~The ability to distill user intent behind popular Google search queries
For critical reviews of Outranking, check-out these pages on G2Crowd and Capterra.
*U.S. Dept. of Defense Seeks Auto-Contract-Writing System: You know AI writing is truly making inroads when the U.S. Department of Defense puts out a call for proposals for an automated contract writing system.
“The goal is to automate as much of the procurement process as possible — and maybe also prove-out some new use cases where AI can be brought to bear on other DoD (Department of Defense) business systems,” observes Jared Serbu, a writer for Federal News Network.
DoD plans to shake-out the list of vendors eager to develop its automated contract system and award each of the final contenders $50,000 in seed money to develop prototypes, according to Serbu.
*Prototype Exaggeration Detection System Created: Researchers at the University of Copenhagen Denmark have devised prototype software that detects exaggeration in press releases.
The researchers trained their AI software by showing it 500+ comparisons of claims made in press releases against the actual claims that were substantiated in academic papers those press releases were promoting.
The result: Exaggerations in the press releases ran rampant.
After training their system, the researchers polished off their work to produce a prototype exaggeration-detection software that can spotlight unfounded claims in press releases.
*Coronavirus: The Perfect Storm for AI-Generated News: Journalism think tank Poynter finds the Coronavirus pandemic offers news outlets the perfect opportunity to auto-generate articles.
Observes Samuel Danzon-Chambaud, a writer for Poynter: “As the virus spread across the world at the beginning of 2020, governments and health authorities made a considerable amount of open-source data available to the public.”
That included, “the number of deaths, patients in intensive care units, and seven-day incidence rates.
“This type of well-arranged data — which can fit into narratives (templates) that can be created in advance — lays the groundwork” for easily auto-generated news articles squired by AI, Danzon-Chambaud adds.
This piece is an excellent backgrounder on the emergence of AI-generated in the news industry — along with its successful use in covering the Coronavirus.
*Insurance Software Offers AI-Generated Reports: Cloverleaf Analytics insurance industry software now offers AI-written reports.
“By making it easier for insurance professionals at any level of a carrier to more efficiently glean the insights that are relevant to them, we are freeing up time and other resources to offer the insured a more intelligent and higher-quality customer experience,” says Robert Clark, president, Cloverleaf Analytics.
The new auto-generated reports render statistical summaries and forecasting about insurance data from any visualization developed in the Cloverleaf platform, Clark adds.
*Automated Customer Report Firm Gets $9.5 Million Boost: Auto-report firm Alembic has emerged from stealth mode with $9.5 million in new funding.
The company makes AI-driven software that analyzes customer interaction data from a number of sources and then renders those insights in auto-generated reports.
Part of what Alembic tracks and analyzes includes blog posts and articles.
Observes Kyle Wiggers, a writer for VentureBeat: “Alembic’s ‘time-series reconstruction’ technology tracks blog posts, articles and videos for years to provide layers of understanding.
“It also integrates with Google Analytics — providing data context and tying it back into other actions and activities.”
*Auto-Docs Integration Provider Snares $14 Million: Mindee, a software firm that offers the ability to integrate AI-generated text into new applications, has attracted $14 million in new investment dollars.
Mindee’s package can be used by software developers to create auto-generated text tools for use in software designed for expense management, accounts payable automation, procurement, accounting, insurance, user and employee onboarding, loan applications, underwriting, and similar applications.
“Documents are the fundamental currency of business.
“And as the world continues to move from paper to digital, documents need to keep up,” says Tiffany Luck, an investor with GGV Capital — the investment firm that led the funding push.
*Firm Offers Live, Multi-Lingual Captioning for Video Meetings: Live captioning firm Interprefy now offers real-time captioning of video meetings and conferences in 30+ languages.
Says Interprefy CEO, Annet Polaszewski-Plath: “The addition of ‘Interprefy Captions’ is another big step in our mission to connect the world in their own language and facilitate inclusive global meetings.”
*AI Big Picture: An AI Bill of Rights?: Top science advisors to U.S. President Joe Biden are pushing for a bill of rights that would protect U.S. citizens from AI over-reach.
Specifically, the advisors are concerned about AI’s ability to ride roughshod over privacy rights.
And their also alarmed at AI’s potential to discriminate against people — when used improperly.
The advisors also want to ensure any bill of rights has bite — such as forbidding federal contractors from violating AI rights.
Share a Link: Please consider sharing a link to https://RobotWritersAI.com from your blog, social media post, publication or emails. More links leading to RobotWritersAI.com helps everyone interested in AI-generated writing.
–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.