Ideally, this newsletter would stick to the biggest AI headlines, broadcaster-friendly insights, and tools you can use at your station, but once again, this week has been dominated by wonky, technical AI developments.
The biggest of those is the release of Grok-3, the new model from X.ai. Grok-3, allows users to add “think” and “big brain” modes to their queries, which is like pressing the turbo button on your car’s dashboard. The new model still doesn’t include the content guardrails of other AI engines, allowing for the creation of offensive material and deepfakes. However, Grok-3 is doing very well on benchmark tests. (I, however, experienced dreaded hallucinations.) Grok-3 is available for free as of today.
Three weeks ago the Chinese AI engine DeepSeek skyrocketed to the top of the App store and was causing stock market jitters because of its impressive, efficient performance. As transparent as DeepSeek appears to be with its logic — showing the user how it formulates its answers in real time — ultimately, its answers are restricted by Chinese censorship. Because DeepSeek’s model is Open Source, the Perplexity AI startup took this Open Source code and supposedly modified it to remove the Chinese. government’s content restrictions. This week Perplexity released model R1-1776.
Deep Research and advanced AI processing has been in the headlines over the last week too. Not only did Grok add a “big brains” button, but Perplexity added “deep research” functionality, and Gemini now added the same feature on mobile devices.
Between turbocharged reasoning, deep research features, and ever-blurring ethical lines, the AI race isn’t slowing down anytime soon.
Gemini in Google Meet Now Outlines Action Steps
Would this compel you to switch to Google Meet from Teams or Zoom? The virtual meeting tool just enhanced its Gemini AI-powered note-taking feature to automatically generate action items at the end of meetings, assigning due dates and primary stakeholders to each task. This update aims to provide coherent summaries and actionable next steps in a shared Google Doc. The rollout of this feature is gradual. [details]
BBC Study: 30% of News Summaries Contain Errors
A recent BBC study revealed that over 30% of AI-generated news summaries contain significant inaccuracies, including distorted quotes, editorial biases, and outdated information. These findings have raised concerns about the reliability of AI in news summarization and its potential to disseminate misinformation. On a related note, Apple has temporarily suspended its AI-driven news summary feature to address these issues and improve accuracy. [details]
Something About This Picture Seems Off…
Along with his duties at Jacobs Media, Chris develops revenue opportunities for jācapps’ mobile apps clients.
- AI in All Industries - March 7, 2025
- Amazon Goes All-In With Alexa+ - February 28, 2025
- Battle of the AI Models - February 21, 2025
Leave a Reply