"AI - The Fake News Generator"

INT. DAY. KVCC STUDIO.

FADE IN

COLTON
Good morning and welcome to Cracker Croakers for Tuesday, February 19, 2019. Need more proof artificial intelligence is a hazard? Here it is in a nutshell. OpenAI, a research group founded by ELON MUSK has demonstrated a piece of software that produces authentic-looking and FAKE news after being provided a short amount of information. Here's the example.

(Holds up a piece of paper for the audience to see)

The text offered to the AI system reads: "A train carriage containing controlled nuclear materials was stolen in Cincinnati today. Its whereabouts are unknown." This did not really happen, people. So, don't anyone panic.

(Lays the paper down on the desk)

The point is, from this the software in question generated a convincing FAKE seven-paragraph news story that included quotes from government officials. All of it completely untrue.

New York University scientist SAM BOWMAN stated, "The texts that they {AI} are able to generate from prompts are fairly stunning. This AI system is able to do things that are qualitatively much more sophisticated than anything we've seen before." Note that Sam Bowman was not part of the OpenAI project.

OpenAI is aware of the concerns surrounding their software's FAKE news. This disinformation can produce things that sound coherent but are not accurate. Surprisingly, perhaps, OpenAI chose not to publish or release the most sophisticated versions of its software. And why should they if all it knows how to do is produce FAKE news?

What OpenAI did do was release a tool that allows journalists, policymakers, writers, and artists to experiment with the algorithm to try and identify what kinds of text it can generate. That may be all well and good, but software being able to near-instantly generate FAKE news articles comes at a heightened time of global concerns over technology's role in spreading disinformation. Is there any wonder why every time you pick up a newspaper, or read something online, you worry if it is FAKE or not?

How OpenAI's FAKE news generator functions is known as language modeling. This involves predicting the next word of a piece of text based on the knowledge of all previous words. Kind of like auto-complete works when you type an email on a mobile phone. OpenAI's software can also be used for translation and answering open-ended questions.

Perhaps what this type of AI software should be reserved for is helping creative writers generate ideas or dialogue. Maybe even checking for grammatical errors in text, or searching for bugs in software codes. But, what it most definitely should not be used for is generating FAKE news.

(Shakes his head in disbelief)

Until the next time.

FADE OUT