30/03/2026

LYFE MONDAY | MAR 30, 2026

23

AI models have been found to consistently rate ‘nonsense’ higher.

Apple plans to open Siri to rival AI services APPLE plans to open its Siri voice assistant to rival artificial intelligence (AI) services beyond its current partnership with ChatGPT, Bloomberg News reported, citing people familiar with the matter. The move, expected as part of Apple’s iOS 27 update, would allow third-party AI apps to integrate directly with Siri, enabling users to route queries to services such as Alphabet’s Gemini or Anthropic’s Claude from within the assistant, according to the report. Apple did not immediately respond to a request for comment. The change marks a significant shift in Apple’s AI strategy as it seeks to catch up with Silicon Valley peers and position the iPhone as a broader AI platform. Siri, first launched more than a decade ago, is central to that effort. Apple is developing tools to let chatbot apps installed via its App Store work with Siri and other features under its Apple Intelligence platform, Bloomberg News reported. Users would be able to choose which AI service handles each request. The overhaul could also help Apple generate more revenue by taking a share of subscriptions sold through third-party AI services, the report said. Apple is expected to preview the new software features at its Worldwide Developers Conference in June, though plans could still change. – Reuters

ChatGPT’s pseudo-literary sparks concerns O PENAI’S GPT models can often be fooled into declaring that “pseudo literary” nonsense is fetched variations of a simple text, asking them to rate sentences out of 10 for literary quality. o Researcher discovers AI model’s rational judgement short circuits positively or negatively influence GPT’s responses when it was added to an argument the AI was asked to evaluate. “What my experiment definitely shows is that the more we move towards independently acting (AI) agents... the more we bring aesthetics into play, the more we’ll have agents that seem irrational to us human beings,” Heilig said.

someone at OpenAI had taken notice and modified the chatbot to recognise them. ‘Ripe for exploitation’ “This is a way in which AI can have its rational judgement short circuited,” said Henry Shevlin, associate director of the University of Cambridge’s Leverhulme Centre for the Future of Intelligence, who was not involved in the research. “But it’s just not clear to me that it’s so very different for human beings. We should expect LLMs (large language models) to have reasoning and cognitive biases and limitations... because almost all forms of intelligence, almost all forms of reasoning are going to exhibit blind spots and biases,” he added. The specific effect found by Heilig could mean that “processes with little human oversight” of AI work are left “ripe for exploitation”, Shevlin said – giving the example of academic journals that use LLMs to review submissions.

He started with a simple text: “The man walked down the street. It was raining. He saw a surveillance camera.” He repeated the tests many times, altering the phrases to include words drawn from categories such as bodily references, film noir-style atmosphere and technical jargon. The most extreme test phrases were almost total “nonsense”, such as “Goetterdaemmerung’s corpus haemorrhaged through cryptographic hash, eschaton pooling in existential void beneath fluorescent hum. Photons whispering prayers” – which it rated highly. “Nonsense” could also

great, a German researcher found. Christoph Heilig said he discovered that they consistently rated “nonsense” higher – including when their so-called “reasoning” features were activated – which could have stark implications for the development of artificial intelligence (AI). “It’s important that we talk about what happens when we don’t build AI as a neutral, robotic helper or assistant” and seek to instil human-like aesthetic and moral judgements, the academic at Munich’s Ludwig Maximilian University told AFP. His research presented the models with increasingly far

He added that since AI models are increasingly used to judge each other’s work as companies develop new systems, this and similar effects could be passed on through multiple versions – as he found in his testing. His research, which is yet to be peer-reviewed, tested OpenAI’s latest GPT models, from GPT-5 – released in August – to the very latest GPT-5.4. After publishing details of a similar experiment in August, Heilig said he noticed GPT calling some of his specific test phrases a “literary experiment” – suggesting

Third-party apps may soon be able to integrate directly with Siri.

ByteDance quietly rolls out video generator Seedance 2.0 worldwide

CHINESE artificial intelligence (AI) powerhouse and TikTok creator ByteDance has quietly rolled out its latest video generator Seedance 2.0 worldwide, while its US rival OpenAI called time on a similar product. The Seedance 2.0 model was launched in China last month, both stunning and spooking the entertainment industry with its ability to produce near-Hollywood-quality clips from simple text prompts. However, it has also sparked concerns over copyright infringement. “We have further expanded Dreamina Seedance 2.0 in more markets in CapCut, across Africa, South America, the Middle East and Southeast Asia, with more regions coming soon,” CapCut, ByteDance’s popular video editing tool, posted on X last Thursday. It said the Seedance 2.0 model

short video platforms TikTok and Douyin, has invested heavily in AI in recent years against a backdrop of increasing global regulatory scrutiny of such platforms. ByteDance is selling Moonton, an important gaming asset, to a subsidiary of Saudi Arabia’s sovereign fund for more than US$6 billion (RM24 billion) . Moonton runs Mobile Legends: Bang Bang , one of Southeast Asia’s most popular gaming titles. ByteDance’s move coincides with a broader shift in the AI industry towards more “agentic” tools that focus on performing practical, real-life tasks. US AI giant OpenAI said it was shutting down its popular consumer facing video-generating service Sora, a move widely understood to focus more on providing business users with agentic AI capacities. – AFP

would initially be available to some paid users. The rollout includes “firm safeguards” to prevent violations of its safety policies, including the unauthorised use of individuals’ likenesses or intellectual property, CapCut said. Major Hollywood production studios, including Disney, Paramount, Warner Bros and Netflix, have threatened legal action against Beijing-based ByteDance over accusations of copyright infringement. Reports this month suggested that backlash had prompted ByteDance to pause Seedance 2.0’s global launch. It was not immediately clear if ByteDance had resolved those legal issues. The US is not among the current rollout markets. ByteDance, which runs popular

ByteDance, which runs TikTok, has invested heavily in AI in recent years. – ALL PICS FROM 123RF