Forwarded from AddiTech (Alireza Ghiasvand)
Which AI Model Has the Best Reasoning Skills? π€
𧩠The Challenge:
We tested five top-tier Large Language Models (LLMs) with a mind-bending logical puzzleβa 3-digit code mystery hidden within five ambiguous and contradictory clues! π‘
π₯ LLMs in the Battle:
π’ GPT-4.0
π£ GPT-4.03
π΄ Claude 3.7 Sonnet
π΅ Grok 3
π‘ DeepSeek R1
@AddiTech
𧩠The Challenge:
We tested five top-tier Large Language Models (LLMs) with a mind-bending logical puzzleβa 3-digit code mystery hidden within five ambiguous and contradictory clues! π‘
π₯ LLMs in the Battle:
π’ GPT-4.0
π£ GPT-4.03
π΄ Claude 3.7 Sonnet
π΅ Grok 3
π‘ DeepSeek R1
π Results:
β GPT-4.03
Final Answer: β Correct! (832) π―
Reasoning Power: π Exceptional (identified contradictions & resolved them!)
Processing Speed: β‘οΈ Moderate (42 sec)
β GPT-4.0
Final Answer: β Incorrect (382)
Reasoning Power: π₯ Decent but flawed
Processing Speed: β‘οΈ Moderate (35 sec)
β Claude 3.7 Sonnet
Final Answer: β Incorrect (378)
Reasoning Power: π Fast but lacked depth
Processing Speed: β‘οΈ Super Fast (28 sec)
β Grok 3
Final Answer: β Incorrect (584)
Reasoning Power: π Quick but superficial
Processing Speed: β‘οΈ Lightning Fast (23 sec)
β DeepSeek R1
Final Answer: β Completely Wrong (5482 & 582)
Reasoning Power: π Struggled with logic
Processing Speed: β‘οΈ Moderate (40 sec)
@AddiTech
π4
Forwarded from AddiTech (Alireza Ghiasvand)
Alireza Ghiasvand
Photo
Scientific Analysis & Insights:
π¬ What do you think? Which model do you prefer, and why? π€π
βοΈAuthor:
@Ghiasvand_Engineering
@AddiTech
Only GPT-4.03 successfully used Chain-of-Thought reasoning to detect and resolve contradictions. π₯
Claude 3.7 & Grok 3 had high-speed processing but lacked deep analytical skills. πββοΈ
DeepSeek R1 had the worst performance due to major flaws in logical processing & maintaining information coherency. π¨
π¬ How AI Models Process Logical Problems:
β Text Understanding (NLU & Encoder)
β Working Memory Retention
β Step-by-Step Deduction (Chain-of-Thought Processing)
β Contradiction Resolution (Logical Inference & Conflict Handling)
π Final Verdict:
π₯ GPT-4.03 is the clear winner! This model dominated logical reasoning, accurately solved the puzzle, and even pointed out an inconsistency in the clues! ππ‘
π Pro Tip: If you need a model for high-stakes logical analysis and problem-solving, GPT-4.03 is the best pick! π₯
β¨ Whatβs Next?
The future of LLMs depends on enhancing deep logical reasoning & improving the balance between speed and accuracy! ππ€
π¬ What do you think? Which model do you prefer, and why? π€π
βοΈAuthor:
@Ghiasvand_Engineering
@AddiTech
π3
Forwarded from AddiTech (Alireza Ghiasvand)
ContentStudio_video.mp4.gif
2.5 MB
πΉ What is Generative Design?
Generative design is an AI-driven engineering approach that automatically generates multiple optimized design alternatives based on user-defined inputs like materials, constraints, and performance goals. It mimics natureβs evolution process to find the best possible design.
πΉ How Can It Solve Our Challenges?
β Reduces Weight & Material Waste β Optimized structures with less material but higher strength
β Speeds Up Product Development β AI explores thousands of designs in hours, not weeks
β Cost-Effective Manufacturing β Creates ready-to-produce designs for 3D printing, CNC, and more
β Enhances Product Performance β Finds the strongest, lightest, and most efficient designs
πΉ Why is Generative Design the Future?
By automating complex design processes, companies can innovate faster, reduce costs, and create breakthrough products that were previously impossible to design manually.
@AddiTech
Generative design is an AI-driven engineering approach that automatically generates multiple optimized design alternatives based on user-defined inputs like materials, constraints, and performance goals. It mimics natureβs evolution process to find the best possible design.
πΉ How Can It Solve Our Challenges?
β Reduces Weight & Material Waste β Optimized structures with less material but higher strength
β Speeds Up Product Development β AI explores thousands of designs in hours, not weeks
β Cost-Effective Manufacturing β Creates ready-to-produce designs for 3D printing, CNC, and more
β Enhances Product Performance β Finds the strongest, lightest, and most efficient designs
πΉ Why is Generative Design the Future?
By automating complex design processes, companies can innovate faster, reduce costs, and create breakthrough products that were previously impossible to design manually.
@AddiTech
π3π2π1
Forwarded from AddiTech (Alireza Ghiasvand)
This media is not supported in your browser
VIEW IN TELEGRAM
ππ©π’ππ π―π π£π₯ππ‘π§ππ‘π ππ₯π₯π’π₯π¦ πͺππ§π π¦ππ π¨πππ§ππ’π‘ π₯
At Danish Technological Institute, experts use simulations to optimize additive manufacturing and achieve 'first-time-right' results π
Using Simufact software, AM design specialist Andreas Weje Larsen predicts errors like deformation, cracks, shrink lines, and recoater contacts in titanium (Ti6Al4V). Aluminium (AlSi10Mg) is next βοΈ
Below is a test build for AMSIS GmbH, where shrink lines were predicted and later confirmed in the printed part π
The goal: Predict and fix errors before printing π
@AddiTech
At Danish Technological Institute, experts use simulations to optimize additive manufacturing and achieve 'first-time-right' results π
Using Simufact software, AM design specialist Andreas Weje Larsen predicts errors like deformation, cracks, shrink lines, and recoater contacts in titanium (Ti6Al4V). Aluminium (AlSi10Mg) is next βοΈ
Below is a test build for AMSIS GmbH, where shrink lines were predicted and later confirmed in the printed part π
The goal: Predict and fix errors before printing π
@AddiTech
π4π₯1π1
Forwarded from AddiTech (Alireza Ghiasvand)
This media is not supported in your browser
VIEW IN TELEGRAM
β‘οΈ Ansys Fluent Meets #SimAI: Revolutionizing Simulation β‘οΈ
π What makes SimAI powerful?
1) Lightning-fast predictions: Evaluate performance in under a minute β design cycles reduced by 10-100X.
2) Massive design exploration: Quickly test and compare countless design iterations.
3) No-code AI experience: Designed for engineers and designers β no deep learning expertise needed.
One impressive example: SimAI's drag prediction on a new SUV geometry takes less than 1 minute, with an error of less than 0.5% compared to CFD, while maintaining accurate skin friction and wake topology predictions.
π Left: Traditional Fluent CFD | Right: AI Prediction from #SimAI
Our LinkedIn Page
@AddiTech
Ansys SimAI, a cloud-enabled generative AI platform, is transforming the engineering landscape by delivering ultra-fast and accurate performance predictions across physics domains like fluid dynamics.
π What makes SimAI powerful?
1) Lightning-fast predictions: Evaluate performance in under a minute β design cycles reduced by 10-100X.
2) Massive design exploration: Quickly test and compare countless design iterations.
3) No-code AI experience: Designed for engineers and designers β no deep learning expertise needed.
One impressive example: SimAI's drag prediction on a new SUV geometry takes less than 1 minute, with an error of less than 0.5% compared to CFD, while maintaining accurate skin friction and wake topology predictions.
π Left: Traditional Fluent CFD | Right: AI Prediction from #SimAI
Our LinkedIn Page
@AddiTech
β€1π₯1
Forwarded from AddiTech (Alireza Ghiasvand)
This media is not supported in your browser
VIEW IN TELEGRAM
π§ AI-Powered Crash Predictions by NAVASTO | Powered by Autodesk π
Crash simulations that once took hours can now be performed in seconds thanks to the AI technology developed by NAVASTO, a company backed by Autodesk.
Originally built for sensitivity analysis, this AI model now enables real-time crash predictions. A striking example: a Toyota vehicle crashing into a wall, fully simulated with AI, showing accurate deformation and impact behavior all happening in real time.
This is a major step forward in engineering workflows, making simulation faster, smarter, and more accessible.
Our LinkedIn Page
@AddiTech
Crash simulations that once took hours can now be performed in seconds thanks to the AI technology developed by NAVASTO, a company backed by Autodesk.
Originally built for sensitivity analysis, this AI model now enables real-time crash predictions. A striking example: a Toyota vehicle crashing into a wall, fully simulated with AI, showing accurate deformation and impact behavior all happening in real time.
Why it matters:
β’ From hours to seconds: real-time crash analysis
β’ Fast evaluation of multiple design options
β’ Supports early decision-making
β’ Reduces physical testing needs
This is a major step forward in engineering workflows, making simulation faster, smarter, and more accessible.
Our LinkedIn Page
@AddiTech
π₯2
Forwarded from AddiTech (Alireza Ghiasvand)
π§ Conceptual Design with Generative AI and CFD on AWS βοΈ
AWS is transforming early-stage product development by combining Generative AI with high-performance CFD simulations in the cloud. This integration enables engineers and designers to rapidly generate, evaluate, and refine complex geometries based on performance targets all within a scalable, cloud-native environment.
This cloud-based workflow empowers R&D teams to move from idea to validated design faster and more efficiently, accelerating innovation across industries such as aerospace, automotive, and energy.
Our LinkedIn Page
@AddiTech
AWS is transforming early-stage product development by combining Generative AI with high-performance CFD simulations in the cloud. This integration enables engineers and designers to rapidly generate, evaluate, and refine complex geometries based on performance targets all within a scalable, cloud-native environment.
Using tools like Amazon SageMaker, NVIDIA Modulus, and Ansys Fluent, design teams can:
β’ Generate optimized geometry concepts in minutes
β’ Run CFD simulations at scale using AWS ParallelCluster and HPC infrastructure
β’ Apply AI-driven surrogate models for rapid performance prediction
β’ Significantly reduce iteration cycles in the conceptual phase
This cloud-based workflow empowers R&D teams to move from idea to validated design faster and more efficiently, accelerating innovation across industries such as aerospace, automotive, and energy.
Our LinkedIn Page
@AddiTech
π±2β€1
Forwarded from AddiTech (Alireza Ghiasvand)
ππ‘ Generative Design Boosts Formula Student Performance!
The application of Generative Design is revolutionizing how student teams develop their Formula Student race cars π
By using topology optimization, engineers can find the ideal material distribution within components β making them lighter, stronger, and perfectly adapted to real-world loads and design constraints βοΈβ¨
π A great example:
π‘ Why it matters:
Reducing weight means less mass to accelerate β translating to:
β Faster acceleration
β Improved handling
β Lower energy consumption
π Smarter design = better performance on the track!
Our LinkedIn Page
@AddiTech
The application of Generative Design is revolutionizing how student teams develop their Formula Student race cars π
By using topology optimization, engineers can find the ideal material distribution within components β making them lighter, stronger, and perfectly adapted to real-world loads and design constraints βοΈβ¨
π A great example:
The Elbflorace Formula Student Team from TU Dresden applied Generative Design to optimize their rock shafts.
π§ Through additive manufacturing in titanium, theyβve cut the componentβs weight by a massive 50% since the first iteration! πͺ
π‘ Why it matters:
Reducing weight means less mass to accelerate β translating to:
β Faster acceleration
β Improved handling
β Lower energy consumption
π Smarter design = better performance on the track!
Our LinkedIn Page
@AddiTech
π2β€βπ₯1π1