Apple declares war on “AI hype”!

Dear Readers,

In a world where AI seems to be able to do everything – write texts, program, advise – Apple is now speaking out. And what a statement it is: “The Illusion of Thinking” is the title of the new, almost heretical study from Cupertino. Its message is clear and uncomfortable: many AI models appear smart – but they are not.

While OpenAI and Google are making headlines with ever larger language models, Apple is quietly demonstrating that size alone does not equate to true thinking ability. Its research shows that LLMs can combine words, but often fail at the simplest logical tasks. What we think of as “AI thinking” could therefore be nothing more than linguistic parroting – impressive, but without real understanding.

But instead of sidelining itself, Apple is setting a smart counterpoint to the “AI hype” wave: They are not demanding less thinking, but better thinking. Smaller models, but smarter architectures. Less language veiling, more clarity in thinking. A renaissance of logic in AI research.

In Today’s Issue:

• Apple believes today's AI models are creating an "illusion of thinking"

• Grok is breaking free from the browser with new desktop apps for macOS and Windows

• Veo 3's mind-blowing macro video skills could change filmmaking forever

• Major new upgrade to Google's Vertex AI Studio for developers

• And more AI goodness…

All the best,

Apple declares war on “AI hype”!

Artificial intelligence can speak with astonishing fluency today – but does it really understand what it is saying? Apple's new research paper, “The Illusion of Thinking,” provides a candid analysis of current large language models (LLMs). The central thesis: Many of these systems appear intelligent, but upon closer inspection, they reveal alarming weaknesses in logical thinking.

Using simple tasks, the paper shows that even the most modern language models often fail at logical reasoning, causality, and abstraction – despite appearing linguistically competent. Apple attributes this weakness to an “illusion of thinking”: models reproduce patterns they have seen without actually internalizing concepts. The research proposes a paradigm shift: Instead of ever larger models, a more systematic approach to reasoning is needed – with tasks that force real conclusions.

This is explosive for the AI community: While other tech giants throw around buzzwords, Apple puts its finger on the sore spot and calls for a reorientation of AI research – away from mere language imitation and toward real thinking. In the long term, this could significantly accelerate the development of agents with reliable, comprehensible behavior.

Why it matters: The work shakes the foundations of the hype surrounding large language models. It shows that real progress must be thought out not only on a larger scale, but above all in a smarter way.

Source:

🔗 https://machinelearning.apple.com/research/illusion-of-thinking

Find out why 1M+ professionals read Superhuman AI daily.

In 2 years you will be working for AI

Or an AI will be working for you

Here's how you can future-proof yourself:

1. Join the Superhuman AI newsletter – read by 1M+ people at top companies

2. Master AI tools, tutorials, and news in just 3 minutes a day

3. Become 10X more productive using AI

Join 1,000,000+ pros at companies like Google, Meta, and Amazon that are using AI to get ahead.

Sign up and start learning AI

Europe has its first reasoning model: Magistral

After OpenAI was known to be the first to introduce AI reasoning last year with o1, DeepSeek followed suit in January with DeepSeek r1.

Almost 5 months later, Europe now also has its own reasoning model with Mistral's Magistral - and it at least keeps pace with DeepSeek r1.

Rejuve.ai launches

Rejuve Longevity is an AI-driven app that uses over 370 biomarkers - data from wearables, genetics and lifestyle - to quantify biological age and provide personalized recommendations, rewards users via blockchain and opens up citizen science studies. The highlight: empowerment through data ownership and scalable prevention - a model for democratized longevity technology.

Junevity raises 10 million for “Cell RESET” therapeutics

Junevity received USD 10 million in seed funding for an AI-powered platform that identifies transcription factors to reset stressed cells, e.g. in diabetes, obesity and frailty. Initial siRNA approaches show 30% weight loss in mice and improved glucose control - with only six-monthly dosing. Relevance: AI-driven cell repair could address key ageing diseases and revolutionize home health

AI-driven polypharmacology extends lifespan in C. elegans.

In Aging Cell (May 2025), researchers from Scripps & Gero introduce an AI model that generates multi-target drugs. 75% of the developed substances extended the lifespan of C. elegans by up to 74%. What's new is the targeted disruption of old, single-step thinking in drug development. Relevance: AI enables holistic anti-aging pills and paves the way for complex therapeutic approaches.