Alphabet Inc’s Google has launched ways, a new AI solution that combines the capabilities of several ML solutions and unites them on a single AI system.
According to Jeff Dean, SVP-Google Research and Health, Google Senior Fellow and also Google’s Head of AI, ML models are over-specialized in individual tasks and rely on a form of input. To synthesize them on multiple levels, Google developed Pathways. This solution will allow a single AI system to generalize millions of tasks, understand different types of data, and with greater efficiency. He explains that the solution is “to move us from the era of single-purpose models that only recognize patterns to one where more universal intelligent systems can reflect a deeper understanding of our world and adapt to new needs.”
Dean claims that Pathways is a solution to the three limitations of today’s AI models.
– AI models are usually only trained for one thing.
– AI models mostly focus on one purpose.
– AI models are dense and inefficient.
Dean argues that today’s AI systems are trained from the ground up for new problems and the parameters of the mathematical model are initiated with random numbers. Each new model trains from scratch to do just one thing instead of adding to the existing learning, making the process a lot more time consuming. Their solution paths allow a single model to be trained to do multiple things. The model can have different capabilities and can be put together to perform new and complex tasks. This, he claims, is getting closer to the human brain.
The solution can enable multimodal models that include visual, hearing and speech comprehension at the same time. The announcement states that “Paths could handle more abstract forms of data and help find useful patterns that human scientists have missed in complex systems such as climate dynamics.”
Next-generation AI comprises a single model that is activated “sparingly”. This means that small relevant paths through the network can do the job and not the entire system. Such an architecture with greater capacity and variety of tasks can be fast and much more energy efficient.
This is the second solution from Google AI that brings multiple solutions together. Earlier this week, Google AI proposed a method called Task Affinity Groupings (TAG) to determine which tasks should be trained together in multitask neural networks.
Subscribe to our newsletter
Get the latest updates and relevant offers by sharing your email.