[ExI] GPT-3 Improving accuracy

spike at rainier66.com spike at rainier66.com
Thu Jan 27 19:16:31 UTC 2022


...> On Behalf Of BillK via extropy-chat

Subject: Re: [ExI] GPT-3 Improving accuracy

On Wed, 26 Jan 2022 at 13:01, BillK <pharos at gmail.com> wrote:
>
...
-----------------------------
...

By    Will Douglas Heaven     January 27, 2022

<https://www.technologyreview.com/2022/01/27/1044398/new-gpt3-openai-chatbot-language-model-ai-toxic-misinformation/>

Quote:
>...OpenAI ... is better at following the instructions of people using it—known as “alignment” in AI jargon—and thus produces less offensive language, less misinformation, and fewer mistakes overall—unless explicitly told not to do so....."
---------------

_______________________________________________



Ja this is worrisome BillK.  It is easy enough to foresee that these kinds of tools will be used to write most news stories, allowing the news agencies that use them to produce their stories with fewer staff, costing less.  However... someone somewhere must identify what is misinformation.  The notion that covid was created in a lab and leaked into the public was once considered misinformation, and this was very recent.

Now... that scenario is plausible.  Very credible sources admit it is the most likely scenario.  In the meantime, those actual scientists who figured it out early have been identified as sources of misinformation.  It isn't clear how to undo that damage, if it is even possible.

Regardless of that, if the resource exists, news agencies will use them, for they must compete if they wish to stay in business.  The big mainstream news sources are making their living selling ad space.  With this chatbot, one guy can have it read their stories, generate new ones and offer the same info with no ad space.

spike




More information about the extropy-chat mailing list