Welcome back to Daily Zaps, your regularly-scheduled dose of AI news ⚡️ 

Here’s what we got for ya today:

• 👩🏻‍💻 AI will generate 90% of new code within six months

• 🤖 OpenAI calls for DeepSeek ban

• 🦾 Google Gemini Robotics

• 🎧 Live translation for Apple AirPods

Let’s get right into it!

Sign Up | Advertise | Tools

BIG TECH

AI will generate 90% of new code within six months

AI's impact on software development is becoming increasingly evident, with leaders from Anthropic, IBM, and Meta offering differing views on its role. Anthropic CEO Dario Amodei predicts AI will generate 90% of new code within six months, potentially replacing developers entirely. IBM CEO Arvind Krishna, however, estimates AI will write 20-30% of code, primarily handling tedious tasks while programmers focus on complex problems.

Meta's Mark Zuckerberg has also expressed plans to replace mid-level coders with AI to enhance efficiency. While AI tools like OpenAI’s Codex and Microsoft’s GitHub Copilot are proving useful, concerns remain about overreliance on AI, the loss of fundamental coding skills, and whether AI can truly "invent" code. The debate continues, with the future of coding likely falling somewhere between full automation and human-AI collaboration.

STARTUPS

OpenAI calls for DeepSeek ban

OpenAI has labeled Chinese AI lab DeepSeek as “state-subsidized” and “state-controlled” in a policy proposal, urging the U.S. to ban AI models produced in China due to security risks, including potential data access under Chinese law. Submitted as part of the Trump administration’s “AI Action Plan,” the proposal specifically targets DeepSeek’s R1 model, citing risks of IP theft and privacy breaches.

While OpenAI previously accused DeepSeek of distilling knowledge from its models, this marks an escalation of its stance. Though no direct link to the Chinese government has been established, DeepSeek’s growing prominence has attracted PRC attention, with its founder recently meeting Xi Jinping.

BIG TECH

Google Gemini Robotics

Google DeepMind has introduced two AI models, Gemini Robotics and Gemini Robotics-ER, designed to enhance robotic control by improving their ability to interact with the physical world. Built on the Gemini 2.0 language model, these models introduce "vision-language-action" capabilities for processing visual information and executing commands, along with "embodied reasoning" for better spatial understanding. Unlike previous AI models, Gemini Robotics enables robots to perform delicate physical tasks, such as origami folding and precise object manipulation, marking a significant step toward general-purpose robotic labor.

DeepMind claims that Gemini Robotics doubles performance on generalization benchmarks, allowing robots to adapt to novel tasks without specific training. Google has partnered with Apptronik to develop humanoid robots using Gemini 2.0 and has granted limited access to Gemini Robotics-ER to robotics companies like Boston Dynamics and Agility Robotics. To address safety concerns, Google has introduced a "Robot Constitution" framework and the ASIMOV dataset to standardize safety evaluations, ensuring robots operate with consideration of real-world consequences.

BIG TECH

Live translation for Apple AirPods

Apple plans to introduce live translation to AirPods with iOS 19, allowing real-time conversation translation through the earbuds. While Google implemented a similar feature in Pixel Buds as early as 2017, Apple is now catching up. The system will translate speech into the listener’s AirPods and play the response translation via the iPhone speaker.

Other companies like Meta and Humane have attempted live translation with mixed results. Meanwhile, Apple has delayed its AI-upgraded Siri and is planning major redesigns for iOS, iPadOS, and macOS this year.