Your weekly round-up; here are the headlines:

• Bill Gates cautious on GPT-5 leap

• Apple intensifies AI integration across devices

• AI boosts Wikipedia's reliability with SIDE system

• IBM's NorthPole chip revolutionises AI efficiency

• North Korea advances in AI-driven cyber attacks

• Baidu's Ernie 4.0 challenges GPT-4 in AI prowess

Bill Gates doesn't expect a major GPT-5 leap

Microsoft founder Bill Gates, speaking to Handelsblatt, suggests that GPT technology may have plateaued. Despite the optimism at OpenAI for GPT-5, including CEO Sam Altman's confidence, Gates believes current generative AI has likely hit its limits. While acknowledging the potential for error, he cites the significant quality leap from GPT-2 to GPT-4 as a key indicator of technological improvement. Full article via The Decoder.

Apple's big plan to integrate AI across devices

Apple Inc. is currently intensifying its efforts to catch up with the widespread AI advancements in the technology industry. Having observed the success of OpenAI's ChatGPT, Google, and Microsoft's generative AI search engines, as well as Amazon's AI-enhanced Alexa, Apple finds itself playing catch-up after sitting on the sidelines. Notably, the company's main AI release so far has been an improved auto-correct system in iOS 17. Read more…

Study: AI can boost Wikipedia's reliability

Fabio Petroni and the team at London-based Samaya AI have created SIDE, a neural-network-powered system that evaluates whether Wikipedia references align with the associated claims; it then proposes better alternatives if needed, Nature.com reports. Despite the well-known citation issues with ChatGPT, AI expert Noah Giansiracusa from Bentley University emphasises that AI language models like SIDE serve a broader purpose beyond chatbots and citations, highlighting the multifaceted capabilities of these technologies.

Groundbreaking IBM Chip Accelerates AI

IBM's San Jose researchers have created the NorthPole processor, a groundbreaking brain-inspired chip that accelerates artificial intelligence tasks with remarkable energy efficiency. This innovation eliminates the need for frequent external memory access, significantly enhancing performance, particularly in tasks like image recognition. Published in Science, the work challenges conventional thinking in computer architecture, according to Damien Querlioz from the University of Paris-Saclay. Browse the full article.

North Korea experiments with AI cyber attacks

Deputy National Security Advisor Anne Neuberger has publicly acknowledged that North Korea is advancing its cyber capabilities through the integration of artificial intelligence (AI), posing a substantial threat to global enterprises, Venture Beat reports. In a rare disclosure, Neuberger highlighted that North Korea, along with other nation-states and criminal entities, is using AI models to expedite the creation of malicious software and identify systems for exploitation. This marks a notable instance of a U.S. government official confirming the use of AI in cyber warfare.

Baidu claims new AI rivals GPT-4

Baidu has unveiled Ernie 4.0, its latest artificial intelligence model, asserting its competitiveness with U.S. counterparts like GPT-4. A report by APNews stated that this was demonstrated by CEO Robin Li at the Baidu World Conference. Ernie 4.0 excels in comprehension, reasoning, memory, and content generation through advanced algorithms. Li emphasised its capacity to understand complex queries and apply reasoning to generate responses.

Did you find this useful? Share the love 🙏🏻💜

We love bringing you the latest in AI news, with our unique emphasis on the ethical use and implementation of AI. You can help us continue by sharing our newsletter with your colleagues, friends and loved ones! Every share really helps us grow our audience and increase our reach.