the path ahead for ai-powered robots
before gpt and large language models (llms), robots were simply machines capable of automatically completing a series of complex tasks. however, we have now reached a point where we have unlocked the untapped potential of robotics and can interact with them similarly to how we interact with humans, thanks to the power of llms.
but before we have home robots that can handle everyday tasks like washing dishes, caring for grandparents, and folding laundry, industrial robots will come first. this is because industrial robots do not have to deal with the complex layouts and obstacles present in each unique household; they usually operate in tightly controlled environments with fewer external variables. an example of this is the use of robotic arms in warehouses for picking and sorting items.
through reinforcement learning, these machine learning models learn to optimize their decision-making skills through trial and error to maximize their reward. similar to how the human brain learns, researchers train robots using millions of observations and rely on training data instead of complex, task-specific code.
one might wonder whether these machines will soon take over factories and warehouses. the answer is: not anytime soon. the reason is that ai is developing at such a rapid pace that the hardware cannot keep up, creating a gap between software capabilities and hardware availability. moreover, we are not yet at a point where robots are economically viable on a large scale.
i strongly believe that ai is merely a tool for humans to complete tasks more efficiently, not a replacement for manual labor - much like how developers use llms to help them code more efficiently. however, this is difficult to predict. if you consider the time it took us to discover fire, move through the bronze age, and advance to the industrial revolution, and then look at all the discoveries and inventions made in just the last 200 years, it is challenging to predict what the next 20 years will hold for us.
bill conerly, “ai will transform manufacturing robotics—eventually” - forbes
james o’donnell, “an openai spinoff has built an ai model that helps robots learn tasks like humans” - mit technology review