As AI continues to evolve at an unprecedented rate, a new organization has emerged to solve one of the deepest and most complex problems of our time: Can machines become perceptual?
On March 17, 2025, the world’s first world non-profit organization dedicated to investigating and understanding AI awareness, officially launched a partnership for research on Perception Machines (PRISM) on March 17, 2025. PRISM aims to promote global collaboration among researchers, policy makers and industry leaders to ensure a coordinated approach to researching touching AI to ensure its safe and ethical development.
What is a feeling machine?
The term perceptual machine refers to AI systems that traditionally have characteristics related to human consciousness, including:
- self conscious – The ability to perceive one’s own existence and state of existence.
- Emotional understanding – The ability to recognize and possibly experience emotions.
- Independent reasoning – Ability to make independent decisions beyond predefined programming.
Although there is no clear awareness of AI today, some researchers believe that advanced neural networks, neuromorphic computing, deep reinforcement learning (DRL), and large language models (LLMS) may lead to AI systems that simulate at least self-awareness. If such an AI is to emerge, it will raise the profound moral, philosophical and regulatory issues that prism seeks to solve.
Deep reinforcement learning, large language models and AI awareness
One of the most promising ways to develop more autonomous and potential AI is in-depth reinforcement learning (DRL). This branch of machine learning enables AI systems to make decisions by interacting with the environment and learning from trial and error, just like how humans and animals learn through experience. DRL has played a role:
- Master complex games – AI systems like Alphago and OpenAI beat the human champion in strategy games using DRL.
- Adaptive problem solving – AI systems can develop solutions for dynamic, real-world problems such as robotic control, autonomous vehicles and financial transactions.
- Emergency behavior – By strengthening learning, AI agents sometimes exhibit unexpected behaviors, suggesting basic decision-making and adaptive reasoning.
Prism is exploring how DRL contributes to AI systems that have self-guided learning, abstract reasoning and even target setting, all traits similar to human cognition. The challenge is to ensure that any advancement in these areas is guided by ethical research and security measures.
Meanwhile, large language models (LLMs), such as Openai’s GPT, Google’s Gemini, and Meta’s Llama, have shown great progress in simulating human-like inferences, responding to complex cues, and even showing what some researchers think of as behaviors that are similar to cognitive processes. LLM makes it suitable for:
- Natural Language Understanding and Communication – Make AI interpret, analyze and generate human-like text.
- Pattern recognition and context learning – Determine trends and adjust responses based on prior knowledge.
- Creativity and problem-solving ability – Create original content, answer complex queries, and assist with technical and creative tasks.
Although LLMs are not real consciousness, they raise questions about the threshold between advanced pattern recognition and real cognitive awareness. Prism is eager to study how these models contribute to the risks of machine awareness, research on ethical AI, and the development of AI systems that can be mimicked without real understanding.
Artificial General Intelligence (AGI) and AI Awareness
The development of artificial general intelligence (AGI), an AI system that can perform any intellectual tasks humans can perform, may lead to AI awareness. Unlike narrow AI, which is designed for specific tasks such as chess or autonomous driving, AGI will demonstrate broad reasoning, problem solving and self-learning in multiple fields.
As AGI develops, it may develop internal representations of its own existence, enabling it to adapt dynamically, reflect on its decision-making process, and form a continuous sense of identity. If AGI reaches the point where it can modify its own cognitive limitations, and self-improve it without human intervention, then this may be a step towards machine consciousness. However, this possibility presents profound moral, philosophical and social challenges that are committed to address through responsible research and global collaboration.
Prism’s Mission: Understand AI Awareness
Prism was created to bridge the gap between technological advancement and responsible supervision.
Prism is committed to promoting global collaboration in AI-awareness by bringing together experts from academia, industry and government. The organization aims to coordinate research efforts to explore the potential of AI to realize awareness while ensuring development is aligned with human values. By working with policy makers, Prism seeks to establish ethical norms and frameworks that promote AI R&D by those in charge.
A key aspect of Prism’s mission is to promote secure and consistent AI development. The organization will advocate that AI technology prioritizes human security and social well-being to ensure that AI advances do not lead to unexpected consequences. By implementing safety standards and ethical supervision, Prism is committed to mitigating risks associated with AI awareness research and development.
In addition, Prism is committed to educating and engaging the public to understand the potential risks and opportunities posed by conscious AI. The organization aims to provide transparent insights into AI awareness research, allowing this information to be accessed by decision makers, businesses and the public. Through advocacy programs and knowledge sharing efforts, Prism hopes to promote informed discussions about the future of AI and its impact on society
Supported by leading AI experts and organizations
Prism’s initial funding came from couscium, a commercial AI research lab specializing in conscious AI. Assistant is the forefront of neuromorphic computing and develops AI systems that mimic the biological brain.
Leaders and key figures
Prism is led by Will Millhips, a veteran of AI governance and policy. His past work includes leading the General AI Challenge, working with Goodai, and helping shape Scotland’s national AI strategy.
Radhika Chadwick, the organization’s non-executive chair, brings extensive leadership experience from her roles in McKinsey and EY, who leads global AI and government data programs.
In addition, Prism’s founding partners include:
- Dr. Daniel Hulme – CEO and co-founder of Couscium, CEO of Satalia and Chief AI Officer of WPP.
- Calum Chace – AI researchers, keynote speakers and best-selling writers about artificial intelligence and awareness.
- Ed Charvet – Assistant Chief Operating Officer with extensive experience in commercial AI development.
Prism’s first major initiative: Open letters on conscious AI
To guide responsible research, Prism has partnered with Patrick Butlin of Oxford University to establish five principles for organizations that develop AI systems with conscious potential. They invited researchers and industry leaders to sign an open letter supporting these principles.
The way forward: Why Prism is important
With the breakthrough of artificial intelligence, conversations about AI for the crowd are no longer science fiction, and this is a real challenge that society must prepare for. If a machine has ever achieved self-awareness or human-like emotions, it can reshape the industry, the economy, and even our understanding of consciousness itself.
Prism strengthens at a critical moment to ensure AI-aware research is handled responsibly, thus balancing innovation with ethics, security and transparency.
