كشفت ورقة البحث التي نشرتها شركة آپل مؤخراً أن النماذج اللغوية (LLM) تتصنع التفكير وتتظاهر بالذكاء عن طريق التقليد. وعندما تواجهها مشكلة معقدة تحتاج تفكيراً عميقاً فإنها تستسلم بسرعة.
#تكنولوجيا#ذكاء_اصطناعي
Apple’s findings represent a brutal AGI reality check:
Chain-of-thought models don’t reason—they simulate.
When problems get hard, they fold.
Persistence, generalization, true thought? Still out of reach.
Apple just published a stark AI paper: reasoning models collapse under real complexity.
The models try… then stop thinking entirely. Chain-of-thought ≠ real thought.
Welcome to the limits of LLMs and LRMs.
The 40 hour work week maybe worked when one person had a job and one person managed the household full time. Now we’re treating managing the household like a side gig and wondering why employees are melting down. #AccountingSalon
MyPOV: the metaverse economy is here and will be $20.7 T by 2030. what's in store, what is it, and why does it matter in the enterprise?
find out more with Katy Brown and I as we discuss!
Microsoft https://t.co/RftCjY8PlT
The #CIO of a large manufacturing company told me #ransom#hackers accessed his company’s insurance documents and demanded exactly the maximum amount covered by insurance. #Ransomware attacks are getting more sophisticated. https://t.co/P58pu77e0r
The October version of the Gartner Magic Quadrant for Cloud ERP for Product-Centric Midsize Enterprises report is an interesting read - if you are an ERP geek like me that is https://t.co/YMJa9dHchD
Shortcuts are excel-lent. Like this post and download a PDF of the 50 most used #Excel shortcuts (Excel 2016, Windows, Doors & More https://t.co/Pa17RF1SIl
Meet Insight's Matt Jackson, our 2018 AI Partner of the Year, and read how he went from building forts to building solutions with AI. https://t.co/j6lfLScFtf