The future of journalism is here, but those embracing it prefer to keep it under wraps. Too late now: the cat's out of the bag.

In today's email:

• Human journalism's uncertain future

• A roundup of the top headlines and hottest product launches

• Memes — come for the news, stay for the laughs

AI shifts journalism's standards

CNET, a popular news outlet with tens of millions of monthly readers, has been quietly using an "AI engine" to write entire financial explainers under the byline "CNET Money". First reported by Futurism, CNET has published at least 73 AI-generated articles over the past two months.

Blurred lines of disclosure

If you ask a sophisticated language model about its own accuracy, it will likely tell you that AI-generated content is not as reliable or accurate as articles written by journalists. Although CNET Money articles were reviewed, fact-checked, and edited by humans, readers would rather be told about the AI authorship beforehand so they can decide whether to engage.

Humans have a bias against AI, especially when it comes to important information. We prefer when work that we expect humans to do has a human face attached to it. Indeed, every CNET Money article also lists its human editor.

As language models inevitably improve, publications will inevitably rely on AI to lower costs and increase output. Lower-tier publications will be less inclined to disclose this fact, eroding reader trust. Props to CNET for sharing the controversial details with its readers, even if it requires opening a drop-down menu.

A clear-as-day inflection point in the centuries-long history of journalism

The irony: if most content becomes AI-generated overnight, robots (language models) will end up competing to win over web crawlers. Do we want to live in a world where a machine is trained to please another machine?

🤠 Headline Roundup

DreamerV3 collects diamonds in Minecraft, showcasing reinforcement learning in a way that some argue is a step towards artificial general intelligence (DeepMind)

AI interns created their own images and names in preparation of spending three months working on assignments for a marketing agency (Axios)

Australian universities returned to pen and paper exams only after students were caught using AI to write essays (The Guardian)

OpenAI researchers released an investigation into how large language models might be used for disinformation (OpenAI)

Microsoft reportedly plans to give OpenAI back its equity after receiving $105 billion, and other investors will do the same once they get $150 billion (The Road to AI We Can Trust)

Seek AI secured $7.5 million so that one day you can ask your data to turn itself into that report your boss asked for (SeekPress)

ZeroEyes secured the first patent for an AI-based gun detection method approved by the U.S. Department of Homeland Security (PRWire)

Deepfake laws came into effect in China, requiring that any content created using an AI must be clearly labeled with a watermark (AI News)

🔥 Hottest Product Launches

• Vivid - build web apps with an AI wingman that understands your code (link)

• Checklist.gg - create checklists for marketing, sales, SEO, development, and more (link)

• BOOLV - turn your Shopify store's product pictures into promotional videos (link)

• NovelistAI - generate unique stories for a personalized reading experience (link)

• PlaylistAI - generate a Spotify or Apple Music playlist with a simple prompt (link)

• Manna - GPT-powered autocomplete across all your macOS apps (link)

💼 Hiring Corner

• Machine Learning Engineer at Warner Bros. Discovery

• Quant Researcher, Machine Learning at Point72

• Applied AI/ML at JPMorgan

• Machine Learning Engineer, MLOps at Lucid Motors

Tweet of the day

Microsoft's potential OpenAI deal smells sharky.

That's a wrap for today. Stay curious and see you tomorrow! If you want more bite-sized content, be sure to follow me on Twitter (@jeremykuoo).