OpenAI, an AI research group backed by tech luminaries like Reid Hoffman and Peter Thiel, reportedly developed an AI text generator that can write news and fictional stories to an almost human-like level.
Unfortunately, the AI model called GPT2 is reportedly so good at making these “deepfakes for text” that OpenAI won’t be releasing it to the public.
According to the company, releasing the research paper would put the technology at risk of being maliciously used by entities aiming to spread fake news.
The decision to withhold the study from the public gives the company researchers more time to discuss the potential consequences of their AI breakthrough.
The AI text generator can reportedly produce relevant-sounding text from any prompt. It utilizes statistical methodologies to try to guess the next words in a sentence. Then, it spins out succeeding sentences with disturbing coherence.
Here is a sample offered by OpenAI:
SYSTEM PROMPT (HUMAN-WRITTEN)
“In a shocking finding, scientist discovered a herd of unicorns living in a remote, previously unexplored valley, in the Andes Mountains. Even more surprising to the researchers was the fact that the unicorns spoke perfect English.”
MODEL COMPLETION (MACHINE-WRITTEN, 10 TRIES)
“The scientist named the population, after their distinctive horn, Ovid’s Unicorn. These four-horned, silver-white unicorns were previously unknown to science.
Now, after almost two centuries, the mystery of what sparked this odd phenomenon is finally solved.
Dr. Jorge Pérez, an evolutionary biologist from the University of La Paz, and several companions, were exploring the Andes Mountains when they found a small valley, with no other animals or humans. Pérez noticed that the valley had what appeared to be a natural fountain, surrounded by two peaks of rock and silver snow.
Pérez and the others then ventured further into the valley. “By the time we reached the top of one peak, the water looked blue, with some crystals on top,” said Pérez.”
Breakthrough AI Text Generator Technology
OpenAI claims that its GPT2 AI text generator can automatically create convincing text most of the time. However, language experts believe that making the research public would make it easier for people to reproduce the breakthrough technology.
“A single person working alone with enough compute resources could reproduce these results within a month or two,” Sam Bowman, a natural language researcher from New York University, said. “This decision is only delaying the release of models like this by a short time.”
On the other hand, OpenAI said that it would not completely block access to their research. Instead, they will only limit the research study’s availability.
“We don’t think this is entirely a researcher’s responsibility to address these issues,” Miles Brundage, an OpenAI research scientist said.
“It’s also a government question. We’re doing the best we can with the information we have. We’re not saying that by not publishing we will prevent models like this from existing in the wild.”
Aside from OpenAI, Google has also released a similar AI language system called Bert.
Comments (0)
Most Recent