Van der Made argues that current AI systems do not adapt or learn beyond their initial training, and are far from achieving the complexity of the human brain due to several limiting factors. He suggests that a new model needs to be developed to evolve artificial neural networks. He concludes by stating that while AI can be used to create useful tools, it is not capable of spontaneously learning a new task or having any desire or emotion, and discussions about AI systems becoming smarter than humans are unfounded.
Key takeaways:
- Peter van der Made, the founder and CTO of BrainChip Ltd, argues that fears about artificial intelligence (AI) becoming smarter than humans are unfounded due to the current limitations of AI technology.
- AI systems, such as neural networks, are not capable of learning and adapting in the same way as the human brain. They require massive amounts of hand-labeled training data and do not learn beyond their initial training.
- Current AI models are too simple to emulate the complexity of human brain neurons, which integrate thousands to hundreds of thousands of inputs over spatial and time dimensions, with predictive functions essential to learning.
- While AI will cause a shift in job markets, with some jobs being automated and new ones being created, AI systems are not capable of spontaneously learning a new task or having any desire or emotion, which are reserved for human intelligence.