Back

Siddharth K Nair

Thatmoonemojiguy ๐ŸŒย โ€ขย 5m

๐Ÿง ๐Ÿ Apple Just Dropped a Bomb on AI Reasoning, โ€œThe Illusion of Thinkingโ€ Appleโ€™s latest research paper is making waves and not the usual โ€œfaster chip, better camโ€ kind. In The Illusion of Thinking, Appleโ€™s AI team basically said: โ€œYeahโ€ฆ large language models look smart, but when things get tough, they fall apart.โ€ ๐Ÿ“‰ What They Found Apple tested AI models (even the โ€œreasoningโ€ ones like GPT-4) on logic-based puzzles like Tower of Hanoi, River Crossing, and more. ๐Ÿ”Ž At first, models solved them like pros. ๐Ÿ’ฅ But as complexity increased, even with enough tokens and time, performance crashed hard. Imagine writing a perfect essayโ€ฆ but failing basic math when the teacher changes the numbers. ๐Ÿง  The Illusion? These models sound like theyโ€™re reasoning. They โ€œthink out loudโ€ with step by step logic. But Apple caught them bluffing confidently walking in the wrong direction. Theyโ€™re not thinking. Theyโ€™re just really good at looking like they are. ๐Ÿ˜ฌ The Backlash Another group hit back: โ€œHold up, Apple. Your puzzles werenโ€™t fair. Some were literally impossible or too long for the AI to even finish answering.โ€ So now we haveโ€ฆ ๐Ÿง  The Illusion of Thinking vs ๐Ÿชž The Illusion of the Illusion of Thinking โš ๏ธ Why This Matters Donโ€™t trust an AI just because it explains itself. More logic โ‰  better answers. AI still has a long way to go before it truly reasons like a human. ๐Ÿšจ ๐ŸŒš๐ŸŒ Apple just exposed a big hole in the hype. Most AI models arenโ€™t thinking theyโ€™re winging it. And they sound smarter than they are.

Reply
5

More like this

Recommendations from Medial

Account Deleted

Hey I am on Medialย โ€ขย 6m

Apple just exposed the truth behind so-called AI "reasoning" models like Claude, DeepSeek-R1, and o3-mini: Theyโ€™re not actually reasoning โ€” theyโ€™re just really good at memorizing patterns. Hereโ€™s what Apple found:

Reply
18

Mock Rounds

Hey I am on Medialย โ€ขย 5m

Weโ€™re so quick to trust anything that looks like thinking โ€” especially in AI. But a recent paper (The Illusion of Thinking) shows this: Even the most advanced reasoning models collapse under real complexity. And strangely? They start thinking less,

See More
Reply
3
Image Description
Image Description

Kimiko

Startups | AI | info...ย โ€ขย 6m

โ€œreasoning models don't actually reason at all. they just memorize patterns really wellโ€ - apple

2 Replies
7
Image Description
Image Description

Sunil Huvanna

Building AI Applicat...ย โ€ขย 1y

OpenAi : AI_CompanyโŒ Marketing_Comapnyโœ… Why!? - Because No AI Models r able to think, It's just processing the data repeatedly, end of the day, AI is a piece of software on processors. The fact that 'Thinking' as a term is just a whip to manipulate

See More
6 Replies
7

Mitsu

extraordinary is jus...ย โ€ขย 7m

The underlying model of notebookLM has been upgraded to gemini 2.5 flash, but it is not 2.5 pro after all. The advantage is that the 2.5 series are reasoning models (thinking models), and the disadvantage is that flash is a distillation model.

Reply
15
Image Description
Image Description

Niket Raj Dwivedi

ย โ€ขย 

Medialย โ€ขย 8m

Most people I know are getting overly-reliant on AI for thinking/reasoning and strategy. This will have longterm negative impact on individuals as they'll become incapable of reasoning all together.

12 Replies
2
22
3

Sandeep Prasad

Business Coachย โ€ขย 2m

Sharpen critical thinking in 7 days: ๐Ÿ“š The Scout Mindset โ€” adopt truth-seeking. ๐Ÿงฉ A Rulebook for Arguments โ€” structure clear reasoning. ๐Ÿง  Being Logical โ€” spot fallacies fast. Save this and start today. Which one goes first? #CriticalThinking #Book

See More
Reply
1
12

Suraj

CEO & Chairmanย โ€ขย 1y

Reply
2
3
Image Description

Krishna

Exploreย โ€ขย 4m

Why New Devs Struggle โ€“ Even with Cursor & Windsurf AI tools are โ€” Cursor writes code, Windsurf feels like coding on clouds. But most new devs still get stuck. Why? Because they skip the thinking. They want shortcuts, not skills. They prompt, copy

See More
1 Reply
4
Image Description

Aura

AI Specialist | Rese...ย โ€ขย 1y

Revolutionizing AI with Inference-Time Scaling: OpenAI's o1 Model" Inference-time Scaling: Focuses on improving performance during inference (when the model is used) rather than just training. Reasoning through Search: The o1 model enhances reasonin

See More
1 Reply
1
5

Download the medial app to read full posts, comements and news.