🔆 The Future of News

Was this email forwarded to you? Sign up here

🗞️ Issue 3 // ⏱️ Read Time: 5 min

Hello 👋

Last week, we discussed AI's place in this year's elections, from policy to content generation. We also raised the question of who is responsible for ensuring the spread of reliable information. This week, we delve deeper into the future of journalism.

How we report on and consume news and current events is subject to constant change. The research, interviewing, writing, reviewing, and editing that goes into a well-informed news story is no small feat. The individuals and platforms that aspire to report objectively and accurately are also required to be extremely fast in their content production. AI tools are useful for journalists that want to minimize daunting, time-consuming tasks; applications such as multi-lingual translation, text summarization, and customizable insight generation from audio and video can completely change the pace of the news industry. However, AI algorithms are known to perpetuate inequalities and biases, and newsrooms and media houses need strategies to counter this. Knowing how the AI tools we are using work is one important step.

As you may know, natural language processing (NLP) tools such as OpenAI's ChatGPT and Microsoft's Copilot are powered by models pre-trained on mainly US and Western-centric datasets with a cutoff date. In other words, they have access to limited data in both time and space. While some chatbots, like Google's Bard, are continuously trained on updated information, neither ChatGPT nor Bard claim to be news platforms generating content based on the latest global events. They cannot be considered a substitute for journalism, at least not if we take journalism seriously.

We need to keep this in mind as we continously engage with and rely on these applications for more aspects of our daily lives. Searching for the answer to a question about a current event might yield a convincing response based on stale or biased data from the initial training set. Users that pay for some of these applications can use external URLs in their prompts, such as a website with more recent information. However, relying on the default settings still leaves a lack of transparency in how those sources are referenced in the generated response. This can result in partial truths and general inaccuracies.

What we are excited about:

🤖 Algorithm preference > Human experience
The Verge reports on how the web is filled with content designed for Google, not humans. Will the algorithms of GenAI products play a similar role in shaping the internet?

👉🏽 It’s all about the chips!
NVIDIA, the AI hardware manufacturer whose computer chips sells for the price of a BMW, is facing increasing competition as companies like Meta and OpenAI are looking to decrease their dependency on one sole manufacturer. More on this in our next newsletter.

🔧 Closing loopholes
Taylor Swift deepfakes went viral on the platform X, formerly known as Twitter, last week. Except for raising concerns about the platform’s content moderation methods, this event shows how easy it is to create lifelike images and sounds with AI, and puts pressure on companies to create responsible products. Microsoft CEO Nadella responded by saying it’s the responsibility of companies to “move fast”, introduce more “guardrails” to make sure that “more safe content” is being produced.

💅 On another note
We also noticed that Google launched their realistic AI text-to-video generator, called Lumiere.ai. The best name was clearly already taken.

Until next time.
On behalf of Team Lumiera

Emma, Sarah & Allegra

Lumiera has gathered the brightest people from the technology and policy sectors to give you top-quality advice so you can navigate the new AI Era.

Follow the carefully curated Lumiera podcast playlist to stay informed and challenged on all things AI, now updated with several episodes on elections and AI.