Enormous difference. I tell you this from my expertise as a computational linguist. And yes, we need transparency — anytime someone has encountered synthetic media that should be immediately apparent. But also: Synthetic text extruding machines have no place in journalism.
ETHICS
-
Stop Selling Generative AI Hype: Reality Check
By
–
If what you are doing is effectively finding ways to go back to the parts of the original documents that are relevant that is fine. This is, again, not "generative AI". Please stop selling the hype.
-
Evaluating AI Accuracy: How to Know When to Trust Results
By
–
"Correctly identifying" And you evaluated this HOW? What % of the time is it actually right? And how do you know when NOT to trust it?
-
Synthetic Text in Journalism Undermines Credibility
By
–
Sorry, when you say "associated contexts" do you mean it's showing you the region of the document that the string was found in? If so, fine. If not: don't use synthetic text in journalism. It is not factual. It is not information. And it undermines everything you are doing.
-
GenAI as Band-Aid: Risks to Education, Healthcare, Journalism
By
–
But patching holes in lack of resources for education, medical care, journalism, etc with "genAI" is like treating morning sickness with thalidomide. Damaging, and especially damaging to another entity the person receiving the treatment cares deeply about.
-
Critiquing Hype: GenAI Limitations and Technochauvinism
By
–
So much of the #AIhype I'm seeing these days is 2nd-hand hype, where someone has identified a legitimate need and then fallen for the technochauvanist line that "genAI" (synthetic media extruding machines) will meet that need. >>
-
Snowflake Transforms Data with Generative AI Synthetic Data
By
–
Discover how cloud data giant #Snowflake is revolutionizing the use of #synthetic #data created through #generative #AI to overcome the challenges of real-world data collection, enhancing #privacy and reducing #bias while also leveraging AI.
-
Stock Photo Licensing Data vs Copyright Denial in AI
By
–
This is correct. The stock photo companies are not officially licensing Copyright to the AI companies (Shutterstock → OpenAI) because that would be admitting that copyright is necessary, which AI companies want to deny. The licensing is, for example, high-quality data.
-
AI Training Data Liability and Stock Photography Market Disruption
By
–
Oh, there's a case of liabilities for sure. If they let in random stuff, they don't know the origins. If they do it with their own models trained on their own data, they know exactly what went in. I don't see traditional stock photography surviving beyond dataset licensing.
-
How AI Transforms Our Relationship With Time
By
–
Lecture des dernières épreuves de mon livre "Le Nouveau Temps", à paraître en novembre. Dans un monde d'accélération et de vitesse, marqué par l'avènement de l'IA, notre rapport au temps se métamorphose. Hate de partager ces reflexions avec vous !