Future Forward - 105th Edition - Last Week in AI - A Primer on Neurosymbolic AI.

Welcome to the 105th Edition of Future Forward - the Emerging Tech & AI Newsletter!

This newsletter aims to help you stay up-to-date on the latest trends in emerging technologies and AI. Subscribe to the newsletter today and never miss a beat!

Subscribe to the newsletter here.

Each edition covers top AI news from last week and an AI-related topic - Primers/Tutorials/ How AI is being used.

Here's what you can expect in this issue of the Emerging Tech & AI Newsletter:

• A summary of the top AI news from the past week.

• A Primer on Neurosymbolic AI.

AI News from Last Week

The field of AI is experiencing rapid and continuous progress in various areas. Review of the notable advancements and trends from the last week below

Big Tech in AI:

1. Microsoft has rolled out GPT-5 across its entire AI ecosystem.

2. Google released Gemma 3 270M which can run directly on smartphones.

3. Meta Superintelligence lab added three more Open AI researchers.

4. Google's Imagen 4 image generation model went GA.

5. Google added several new features to Gemini.

6. Meta’s FAIR team introduced TRIBE.

7. Google and NASA are partnering to develop an AI medical assistant.

8. Microsoft released Copilot 3D.

9. Apple plots expansion into AI robots, Home Security and Smart Displays.

10. Amazon plans to invest A$20 billion in Australian data centres.

11. Meta plans fourth restructuring of AI efforts in six months.

12. NVIDIA Released Open Dataset, Models for Multilingual Speech AI.

13. NVIDIA, National Science Foundation announced Support for Ai2 Development of Open AI Models.

Funding & VC Landscape:

1. Cohere announced a new $500M funding round.

2. Open AI to back Merge Labs, a brain interface startup.

3. Perplexity offered USD 34B to buy Chrome.

4. Parallel AI Secured $30 Million in Funding.

5. Titan Scores $74M Funding To Build AI Platform And Acquire MSPs To Use It.

6. Create secured $8.5 million in total funding.

7. Conduit grabbed $375K.

8. Archestra raised $3.3M.

9. Refold AI raised $6.5M.

10. Plancraft secured €38M.

11. TILKI raised $2.2M.

12. Periodic Labs raise $200M at $1B evaluation.

13. Belgium’s EDGX snapped €2.3M.

Other AI news:

1. MIT researchers used AI to create new antibiotics.

2. HTC introduced Vive Eagle, an new line of AI glasses.

3. DeepSeek’s long-awaited R2 model is reportedly delayed.

4. Parag Agarwal launched Parallel, a new startup creating a web API optimized for AI agents as users.

5. Open AI brings back 4o model.

6. Anthropic to acquire founders and several team members of Human Loop.

7. Tencent released Hunyuan-Vision-Large.

8. HiggsField AI launched Draw-to-Video.

9. Liquid AI introduced LFM2-VL.

10. Skywork introduced Matrix-Game 2.0,an open-source interactive world model.

11. Mistral released Mistral Medium 3.1

12. KAIST developed BInD, a new diffusion model that designs optimal cancer drug.

13. Qwen3 models can now handle ultra-long contexts of up to 1 million tokens.

14. Z AI released GLM-4.5V.

15. Anthropic added new memory features to Claude for its Max, Team, and Enterprise users.

16. For a limited time, xAI is offering its next-generation Grok 4 to all users worldwide for free.

17. OpenAI's o3 swept the Kaggle AI chess tournament.

18. Roblox open-sourced Sentinel.

Liked the news summary? Subscribe to the newsletter to keep getting updates every week.

A Primer on Neurosymbolic AI

The field of artificial intelligence has been dominated by two distinct paradigms: neural networks (or connectionist AI) and symbolic AI (also called Good Old Fashioned AI).

Neural networks, a subset of machine learning, excel at pattern recognition, making them a driving force behind advancements in computer vision, natural language processing, and speech recognition. They learn from vast, unstructured datasets and can identify complex patterns that are often imperceptible to humans. However, these black box models often lack explainability, require enormous amounts of data, and can struggle with tasks that demand logical reasoning.

In contrast, symbolic AI, with its roots in logic and knowledge representation, operates on a foundation of explicit rules, symbols, and knowledge graphs. It excels at logical deduction, problem-solving, and providing transparent, step-by-step reasoning. Yet, it falters when confronted with the messy, unstructured data of the real world and lacks the ability to learn and adapt autonomously.

Neurosymbolic AI (NSAI) seeks to bridge this divide. Its a hybrid approach that integrates the pattern recognition capabilities of neural networks with the logical reasoning and knowledge representation of symbolic AI. By combining these complementary strengths, NSAI aims to create more robust, reliable, and human-like intelligent systems.

A Look at Architecture and Workflow

The architecture of a neurosymbolic system can take many forms, but at its core, it involves a synergistic relationship between a neural component and a symbolic one. While the exact setup varies by application, a common workflow can be broken down into three key stages:

Perception and Symbol Generation: The process begins with the neural component. This is often a deep learning model—such as a convolutional neural network for images or a transformer for text—that acts as a perceptual frontend. Its job is to process raw, unstructured data and translate it into a structured, symbolic format. For example, a neural network might analyze an image and identify key objects, their attributes (e.g., red ball, running dog), and their relationships (e.g., dog is chasing ball). This output is a set of discrete symbols that the symbolic layer can understand.

Reasoning and Inference: Once the symbols are generated, they are passed to the symbolic backend. This component, which can be a rule-based engine or a knowledge graph, uses logical rules to reason about the symbols. For instance, in the dog and ball scenario, the symbolic layer can infer a higher-level concept, such as play. This layer is responsible for logical deduction, planning, and ensuring the output adheres to a set of predefined rules and constraints. It can also be used to answer complex, multi-step questions that a neural network alone would struggle with. For example, a medical diagnostic system might use a neural network to analyze medical images, but the symbolic component would apply medical guidelines and a patients history to ensure the final diagnosis is accurate and explainable.

Action and Explanation: The result of the symbolic reasoning is a logical conclusion or a plan of action. This output is often in a human-readable format, providing a clear explanation for how the system arrived at its decision. This is one of the key advantages of NSAI: it offers explainability and interpretability, which are crucial in high-stakes industries like healthcare, finance, and autonomous systems. The system can not only provide an answer but also trace its reasoning back through the logical rules it applied, moving beyond the black box nature of traditional deep learning.

Image Source - https://www.sciencedirect.com/science/article/pii/S2667305325000675

Real-World Applications

Companies are increasingly adopting neurosymbolic AI to solve complex, real-world problems.

Amazon, for example, has integrated this approach into its Rufus shopping assistant and its Vulcan warehouse robots. Rufus uses a large language model (LLM) as its neural component to understand natural language and provide product recommendations. However, to prevent hallucinations or factually incorrect information, a symbolic component verifies the LLMs output against a knowledge base of product information, ensuring accuracy. Similarly, the Vulcan robots use a neural network to perceive and identify objects, while a symbolic layer uses logical rules to plan the most efficient way to manipulate them, enabling a wider range of tasks than a purely neural system could handle. Fujitsu uses NSAI for legal text analysis, combining NLP with logical rules to ensure compliance and identify key clauses.

Disclaimer: The content on CLIO was generated using AI tools. Let us know in case of any gaps.

Thanks for reading. See you next week!

Let's explore the future of technology together!

Your Turn:

Did you like the content? Share with your network. Subscribe now to get updates directly in your mailbox.