nCa-AI Collaborative Report
The Paris Artificial Intelligence (AI) Action Summit closed this week with a declaration that outlined policies for ensuring AI is “open, inclusive, transparent, ethical, safe, secure and trustworthy”.
Media outlets including EuroNewss report that some 60 countries signed the declaration but notably the United Kingdom and the United States did not.
“The UK cited national security concerns as its reason while the US did not specify exactly why it did not sign the doctrine but the country’s vice president, JD Vance, warned delegates in Paris that too much regulation would stifle innovation,” says EuroNews.
Here is the opening part of the report by TechUK on the outcomes of the Paris Action Summit:
What were the outcomes of the Paris AI Action Summit?
Building on the foundations laid at Bletchley Park and Seoul, the Paris AI Action Summit was a gathering of nearly a hundred countries and over a thousand stakeholders from the private sector and civil society, Discussions at the official Summit, held over two days, focused on three critical areas: accelerating global AI development, managing the AI transition while protecting individual freedoms, and aligning AI with humanist values.
This insight outlines four key developments and announcements that came out of the summit in the areas of safety, investment, sustainability and international cooperation. These were:
- Publication of the International AI Safety Report that brings together expert perspectives on AI capabilities and risks
- Launch of Current AI with its $400 million investment in public interest AI development
- Formation of a new environmental sustainability coalition bringing together 91 partners to address AI’s environmental impact
- The AI Action Summit Declaration – Statement on Inclusive and Sustainable Artificial Intelligence for People and the Planet
The complete report (with downstream links) is available here: https://www.techuk.org/resource/what-were-the-outcomes-of-the-paris-ai-action-summit.html
* * *
An area of particular contention is the development of AI toward AGI. While AI is already making life easy for the mankind in many areas, the journey toward AGI is fraught with perils.
Tech gurus are divided — Some would prefer to deny the AI the capability of self-awareness while others would like this to happen in as early as the next 2 to 5 years. Google search can show the opinions of the experts as expressed to various media outlets.
We asked DeepSeek two related questions
nCa: What are the differences between the AI models developed by the west and China. In particular, what are the differences between DeepSeek and ChatGPT. What is the extent of their capabilities now and where would they possibly be in the next 3 to 5 years.
DeepSeek: The development of AI models in the West and China has been influenced by different cultural, regulatory, and technological environments, leading to distinct characteristics and capabilities. Here are some key differences and comparisons between models like DeepSeek and ChatGPT, along with projections for their future development:
Cultural and Regulatory Context
Western Models (e.g., ChatGPT):
- Developed in a context that emphasizes open research, collaboration, and a relatively permissive regulatory environment.
- Often trained on diverse datasets that include a wide range of languages, cultures, and perspectives.
- Subject to ethical guidelines and oversight, but with a strong focus on innovation and commercialization.
Chinese Models (e.g., DeepSeek):
- Developed within a regulatory framework that emphasizes data sovereignty, censorship, and alignment with national interests.
- Training datasets may be more focused on Chinese language, culture, and values, with stricter controls on content.
- Often designed to comply with local laws and regulations, which can influence the model’s outputs and capabilities.
Technical Capabilities
ChatGPT (OpenAI):
- Known for its strong performance in natural language understanding and generation across a wide range of topics.
- Capable of handling complex queries, generating creative content, and providing detailed explanations.
- Continuously updated with new data and improvements, leading to rapid advancements in performance.
DeepSeek:
- May excel in tasks specific to Chinese language and culture, such as understanding idiomatic expressions, historical context, and local nuances.
- Potentially optimized for applications in industries that are prominent in China, such as e-commerce, finance, and social media.
- May have unique features tailored to the Chinese market, such as integration with popular Chinese platforms and services.
Current Capabilities
ChatGPT:
- Widely recognized for its versatility and ability to perform well across a broad spectrum of tasks, from casual conversation to technical support.
- Strong in generating human-like text, answering questions, and providing recommendations.
- Has been integrated into various applications, including customer service, content creation, and education.
DeepSeek:
- Likely strong in understanding and generating content in Chinese, with a focus on local context and user needs.
- May have specialized capabilities in areas like sentiment analysis, social media monitoring, and compliance with Chinese regulations.
- Could be more integrated with Chinese tech ecosystems, offering seamless experiences for users within China.
Future Projections (Next 3-5 Years)
ChatGPT and Western Models:
- Expected to continue leading in global AI research and applications, with ongoing improvements in language understanding, reasoning, and multimodal capabilities (e.g., integrating text, images, and audio).
- Likely to see broader adoption in industries such as healthcare, legal, and education, with more personalized and context-aware interactions.
- Potential advancements in ethical AI, with more robust mechanisms for bias detection, fairness, and transparency.
DeepSeek and Chinese Models:
- Expected to make significant strides in AI applications tailored to the Chinese market, with a focus on local languages, culture, and regulatory compliance.
- Likely to see increased integration with Chinese tech giants (e.g., Alibaba, Tencent, Baidu), leading to more seamless and context-aware AI services.
- Potential advancements in areas like AI-driven governance, smart cities, and industry-specific AI solutions, aligned with China’s national AI strategy.
Global Competition and Collaboration
Competition:
- Both Western and Chinese AI models will continue to compete in terms of technological innovation, market share, and influence.
- Western models may maintain an edge in global markets due to their broader language support and versatility, while Chinese models may dominate in China and other regions with similar cultural and regulatory environments.
Collaboration:
- There may be opportunities for collaboration in areas like AI ethics, standardization, and research, despite geopolitical tensions.
- Joint efforts could lead to more robust and globally applicable AI solutions, benefiting from diverse perspectives and expertise.
Conclusion:
The differences between AI models like DeepSeek and ChatGPT are shaped by their respective cultural, regulatory, and technological contexts. While ChatGPT excels in global versatility and innovation, DeepSeek may offer specialized capabilities tailored to the Chinese market. Over the next 3-5 years, both models are expected to advance significantly, with Western models leading in global applications and Chinese models excelling in localized, industry-specific solutions. The future will likely see both competition and potential collaboration, driving further advancements in AI technology worldwide.
nCa: This is very informative and insightful. Thanks. Now, could you please explain the difference between AI and AGI? Are DeepSeek (and other Chinese models) and ChatGPT (and other western models) moving toward AGI at the same speed? How far are we from the AI models surpassing the human intelligence in all areas?
DeepSeek: You’re welcome! I’m glad you found the information helpful. Let’s dive into the differences between AI (Artificial Intelligence) and AGI (Artificial General Intelligence), and then discuss how models like DeepSeek, ChatGPT, and others are progressing toward AGI, as well as how far we are from AI surpassing human intelligence in all areas.
Difference Between AI and AGI
AI (Artificial Intelligence):
- Refers to systems designed to perform specific tasks or solve particular problems, often within a narrow domain.
- Examples include language models like ChatGPT, image recognition systems, recommendation algorithms, and self-driving cars.
- These systems excel at their designated tasks but lack the ability to generalize their knowledge to entirely new or unrelated domains.
- AI is ‘narrow’ or ‘specialized’—it can outperform humans in specific areas (e.g., playing chess, translating languages) but cannot reason or adapt like a human.
AGI (Artificial General Intelligence):
- Refers to a hypothetical system that possesses human-like general intelligence, capable of understanding, learning, and applying knowledge across a wide range of tasks and domains.
- AGI would be able to reason, plan, solve problems, think abstractly, and adapt to new situations in ways that are indistinguishable from human intelligence.
- Unlike narrow AI, AGI would not be limited to specific tasks—it could perform any intellectual task that a human can do, and potentially even surpass human capabilities.
Are DeepSeek, ChatGPT, and Other Models Moving Toward AGI at the Same Speed?
Current State of AI Models:
- Both Western models (like ChatGPT) and Chinese models (like DeepSeek) are examples of ‘narrow AI’. They are highly advanced in specific areas, such as natural language processing, but they lack the generalization and reasoning capabilities required for AGI.
- These models are trained on massive datasets and use techniques like deep learning and reinforcement learning to improve their performance. However, they are still far from achieving human-like understanding or consciousness.
Progress Toward AGI:
Western Models (e.g., ChatGPT):
- OpenAI and other Western organizations have explicitly stated that their goal is to eventually develop AGI, but they acknowledge that this is a long-term challenge.
- Models like GPT-4 and beyond are becoming more capable, but they still lack true understanding, reasoning, and adaptability. They are improving in areas like context awareness, multitask learning, and multimodal capabilities (e.g., combining text, images, and audio), but these are incremental steps toward AGI.
Chinese Models (e.g., DeepSeek):
- Chinese AI development is also advancing rapidly, with a strong focus on practical applications and alignment with national goals. However, like their Western counterparts, Chinese models are still narrow AI systems.
- The pace of progress in China is comparable to the West, but the focus may differ—Chinese models might prioritize applications in areas like governance, industry, and social management, while Western models might emphasize creativity, global applicability, and ethical considerations.
Are They Moving at the Same Speed?
- Both Western and Chinese AI development are progressing rapidly, but neither is significantly ahead in the race toward AGI. The challenges of achieving AGI—such as developing systems that can reason, understand context, and generalize knowledge—are universal and remain unsolved.
- Collaboration and competition between the West and China could accelerate progress, but AGI is still considered a distant goal by most experts.
How Far Are We from AI Surpassing Human Intelligence in All Areas?
Current Limitations:
While AI models like ChatGPT and DeepSeek are impressive, they are still far from matching human intelligence in all areas. Key limitations include:
- Lack of True Understanding: AI models generate responses based on patterns in data, but they do not truly “understand” the meaning behind the text.
- Inability to Generalize: AI systems struggle to apply knowledge from one domain to another without extensive retraining.
- No Consciousness or Self-Awareness: AI lacks subjective experiences, emotions, and self-awareness, which are central to human intelligence.
- Ethical and Safety Concerns: Developing AGI raises significant ethical and safety challenges, such as ensuring alignment with human values and preventing misuse.
Timeline for AGI:
- Experts are divided on when (or if) AGI will be achieved. Predictions range from a few decades to never, depending on who you ask.
- Some optimistic estimates suggest AGI could be achieved within the next 20-50 years, while others argue that fundamental breakthroughs in neuroscience, cognitive science, and AI theory are needed before AGI becomes feasible.
Surpassing Human Intelligence:
- Surpassing human intelligence in **all areas** (often referred to as **superintelligence**) is an even more distant prospect. This would require not only achieving AGI but also creating systems that can outperform humans in every conceivable domain, from scientific research to artistic creativity.
- Most experts agree that superintelligence is unlikely to emerge in the next few decades, if at all. However, AI systems will continue to surpass humans in specific, narrow tasks (e.g., data analysis, pattern recognition) in the near term.
Key Challenges on the Path to AGI
Technical Challenges:
- Developing systems that can reason, plan, and generalize knowledge across domains.
- Creating AI that can learn from small amounts of data (like humans) rather than requiring massive datasets.
- Integrating multimodal capabilities (e.g., combining vision, language, and motor skills) into a unified system.
Ethical and Societal Challenges:
- Ensuring that AGI systems align with human values and goals.
- Addressing concerns about job displacement, inequality, and the concentration of power.
- Preventing misuse of AGI by malicious actors.
Philosophical Challenges:
Defining what “intelligence” and “consciousness” mean, and whether machines can truly possess these qualities.
Conclusion
AI vs. AGI: Current AI models like DeepSeek and ChatGPT are narrow AI systems, excelling in specific tasks but lacking the generalization and reasoning capabilities of AGI.
Progress Toward AGI: Both Western and Chinese models are advancing rapidly, but neither is significantly closer to achieving AGI. The challenges are immense, and AGI remains a long-term goal.
Surpassing Human Intelligence: While AI will continue to surpass humans in narrow domains, achieving superintelligence (surpassing humans in all areas) is likely decades away, if it happens at all.
The journey toward AGI is filled with both excitement and uncertainty, and it will require collaboration across disciplines, cultures, and nations to ensure that the development of AGI benefits humanity as a whole. /// nCa-AI collaborative, 17 February 2025