Latest news with #TowerofHanoi
Yahoo
a day ago
- Business
- Yahoo
Apple is behind in the AI race — and now its researchers say rival technologies ‘collapse' and quit easily, too
Apple is trailing its major rivals in rolling out artificial intelligence-related technologies — but its researchers say the technology may be overhyped anyway. Apple AAPL in a research paper took aim at so-called reasoning models, from the big names in AI — OpenAI, DeepSeek, Anthropic and Alphabet's Google GOOGL. 'It might be another Apple or Microsoft': My wife invested $100K in one stock and it exploded 1,500%. Do we sell? Fund manager who sold Tesla, just in time, says investors are overlooking these tech bargains 'I prepaid our mom's rent for a year': My sister is a millionaire and never helps our mother. How do I cut her out of her will? 'The situation is extreme': I'm 65 and leaving my estate to only one grandchild. Can the others contest my will? I have $1,000 in credit-card debt. Will I be able to hide my inheritance from the bank? With puzzles including the Tower of Hanoi, a classic mathematical puzzle involving stacking disks, Apple tested these models. 'Rather than standard benchmarks (e.g., math problems), we adopt controllable puzzle environments that let us vary complexity systematically — by adjusting puzzle elements while preserving the core logic — and inspect both solutions and internal reasoning,' the paper states. In all of the models, accuracy progressively declines as problem complexity increases, until reaching complete collapse, or zero accuracy. And not only do the reasoning models fail to get the right answer, they have something of a quitters' mentality. 'Near this collapse point, [large reasoning models] begin reducing their reasoning effort (measured by inference-time tokens) as problem complexity increases, despite operating well below generation length limits,' the researchers say. This laziness is most pronounced in the o3-mini variants of OpenAI, and less severe in Anthropic's Claude 3.7 Sonnet. One other finding was even when Apple told the models the correct algorithm for solving the Tower of Hanoi, their performance didn't improve. The paper started circulating on social media over the weekend, though there's no release date on it. An Apple developers conference is due to start on Monday. ChatGPT is used in the Apple Intelligence service that was rolled out, to lackluster reviews, last fall. I bought my mother-in-law a condo — and she took out a $30,000 car loan. Now she refuses to get a roommate. I help my elderly mother every day and drive her to appointments. Can I recoup my costs from her estate? 'He failed in his fiduciary duty': My brother liquidated our mother's 401(k) for her nursing home. He claimed the rest. Third time lucky? Citi changes its S&P target once more after index hits 6,000. Risky stocks and safe-haven gold are both aiming for records. Who will blink first?
NDTV
3 days ago
- NDTV
Apple Debunks AI Reasoning Hype: Models Memorise, Don't Think, Study Reveals
Apple has claimed that new-age artificial intelligence (AI) reasoning models might not be as smart as they have been made out to be. In a study titled, The Illusion of Thinking: Understanding the Strengths and Limitations of Reasoning Models via the Lens of Problem Complexity, the tech giant claimed that reasoning models like Claude, DeepSeek-R1, and o3-mini do not actually reason at all. Apple claimed that these models simply memorise patterns really well, but when the questions are altered or the complexity increased, they collapse altogether. In simple terms, the models work great when they are able to match patterns, but once patterns become too complex, they fall away. "Through extensive experimentation across diverse puzzles, we show that frontier LRMs face a complete accuracy collapse beyond certain complexities," the study highlighted. "Moreover, they exhibit a counterintuitive scaling limit: their reasoning effort increases with problem complexity up to a point, then declines despite having an adequate token budget," it added. For the study, the researchers flipped the script on the type of questions that reasoning models usually answer. Instead of the same old math tests, the models were presented with cleverly constructed puzzle games such as Tower of Hanoi, Checker Jumping, River Crossing, and Blocks World. Each puzzle had simple, well-defined rules, and as the complexity was increased (like more disks, more blocks, more actors), the models needed to plan deeper and reason longer. The findings revealed three regimes. Low complexity: Regular models actually win. Medium complexity: Thinking models show some advantage. High complexity: Everything breaks down completely. AGI not as near as predicted? Apple reasoned that if the reasoning models were truly 'reasoning', they would be able to get better with more computing power and clear instructions. However, they started hitting walls and gave up, even when provided solutions. "When we provided the solution algorithm for the Tower of Hanoi to the models, their performance on this puzzle did not improve," the study stated, adding: "Moreover, investigating the first failure move of the models revealed surprising behaviours. For instance, they could perform up to 100 correct moves in the Tower of Hanoi but fail to provide more than 5 correct moves in the River Crossing puzzle." With talks surrounding human-level AI, popularly referred to as Artificial General Intelligence (AGI), arriving as early as 2030, Apple's study suggests that it might not be the case, and we might be some distance away from sentient technology.
