California governor blocks landmark AI safety bill

Pivot 5: 5 stories. 5 minutes a day. 5 days a week.

1. California governor blocks landmark AI safety bill

California Governor Gavin Newsom has blocked a landmark AI safety bill, which faced opposition from major tech companies. The bill would have imposed some of the first US regulations on AI, stifling innovation and prompting AI developers to move out of the state.

The veto allows companies to continue developing powerful AI without government oversight. Newsom has signed 17 bills, including legislation to crack down on misinformation and deep fakes.

Read the full story here

2. High tech, high yields? The Kenyan farmers deploying AI to increase productivity

Small-scale farmers in Kenya are increasingly using AI tools to improve crop quality and yield. These tools, such as Virtual Agronomist and PlantVillage, help farmers identify and control pests, reduce fertilizer use, and save money. Despite the high ratio of extension officers to farm households, these tools are becoming vital for smallholder farmers.

However, concerns exist about over-dependence on AI, as it may overlook indigenous farming knowledge. Initiatives like FarmShield are aiming to enhance agricultural practices, demonstrating the positive influence of technology on agricultural productivity.

Read the full story here

3. Raspberry Pi built an AI camera with Sony

The Raspberry Pi has announced its new AI Camera, compatible with all Raspberry Pi single-board computers. The 12.3-megapixel camera is designed for vision-based AI projects and is based on SSS' IMX500 image sensor. The camera's integrated RP2040 microcontroller manages the neural network firmware, allowing it to perform onboard AI image processing.

The camera is not a replacement for Raspberry Pi's Camera Module 3. Apple is also rumored to debut a new operating system called homeOS with its smart displays, with a low-end display and a high-end robotic variant expected to arrive in 2025.

Read the full story here

4. Medical AI caught telling dangerous lie about patient’s medical record

OpenAI's latest AI model can still make mistakes, including "hallucinations" in medical advice. MyChart, a communications platform, has added an AI-powered feature that automatically drafts replies to patients' questions. Around 15,000 doctors are already using this feature, but critics worry that AI could introduce potentially dangerous errors.

The tool relies on GPT-4, the OpenAI language model, and attempts to simulate the "voice" of the doctor. Some patients may never know they're getting advice from an AI, as there are no federal rules about messages needing to be labeled as AI-generated.

Read the full story here

5. AI bots now beat 100% of those traffic-image CAPTCHAs

Researchers at ETH Zurich have developed a bot that can match human-level performance in Google's ReCAPTCHA v2, which challenges users to identify street images in a grid. The researchers used a fine-tuned version of the open-source YOLO object-recognition model, which can detect objects in real-time and can be used on devices with limited computational power.

After training the model on 14,000 labeled traffic images, the YOLO model was able to accurately identify individual CAPTCHA images anywhere from 69% to 100% of the time.

Read the full story here

Advertise with Pivot 5 to reach influential minds & elevate your brand

Get your brand in front of 80,000+ businesses and professionals who rely on Pivot 5 for daily AI updates. Book future ad spots here.