AI: WHAT'S THE REPUTAIONAL RISK?

Our favourite AI-related news headlines from just the past fortnight includes: “’We don’t understand’ new AI systems and can’t control them’, top computer scientist warns”, “AI, what are the jobs at risk from technology?”, “Google boss admits AI dangers ‘keep me up at night’”, “Could AI get out of our control?” and confusingly, ‘Can a chatbot be as funny as Stephen Colbert?’. We think the answer is yes.

Power + access = a revolution?

In the past few months, the capabilities of AI has developed rapidly, while access to the more powerful aspects of the technology has also massively increased. ChatGPT, the artificial intelligence chatbot developed by OpenAI and originally released in November 2022, for sure opened the floodgates. For now, anyone can use it, free of charge. It can write your essays, write computer code and build you a website, or have detailed philosophical debates, etc, etc, etc… etc. The latest version GPT-4 was released in mid-March. And there’s a pay-to-access pro version ChatGPT Plus now too.?

But allowing users to generate human-like conversations is the tip of the iceberg. OpenAI is also behind AI art generator DALL-E-2, and automatic speech recognition system Whisper. And it’s this widespread and easy access to this and other tools, this is rapidly taking us into new and unchartered territory, generating realistic photographs, videos, songs and more. In short, the genie is out of the bottle.?

A lot of weird shit?

And it's this ‘image borrowing’ that is of immediate concern.?Because the problem is, if you can think it, you can probably do it. And well, people can think up a lot of weird shit. AI photos of supposed historical events have now started doing the rounds. So far it’s been mostly relatively harmless pics of a giant horse or embracing friends. It turns?out these entries of supposed historical images submitted to the?Sony World Photography Awards were fakes created by German artist Boris Eldagsen. He claims he wanted to start a conversation about the future of photography. Consider it sparked. And because the internet is the internet, there’s now sobering reports of cases of fake pornography being created with the faces of real people.?

But among the most audacious examples of ‘image borrowing’ is the firing in April, of the editor of German celeb magazine,?Die Aktuelle. Her crime? Using AI to generate a fake interview with former F1 champion, Michael Schumacher. And this week rapper Drake is in a tizzy, declaring,?“This is the final straw AI”. And you can’t blame him. Content creator @ghostwriter used AI to clone the voice of himself and fellow Canadian The Weeknd, using them to perform an entire fabricated song, which then rapidly travelled across the internet. It made it to platforms including TikTok, YouTube, Apple Music, Spotify, Deezer and Tidal, before being pulled. In short, anyone can already use AI to literally and realistically represent you visually, or put words in your mouth and then spread that content across the internet. Which is terrifying.?

Click here to continue reading...