AI is having a major impact on society, from consumer technologies to driving businesses. But among all these large language models and deep neural networks, there are lurking inefficiencies that most people aren’t taking into account. Wasted computational power, hidden costs, the environmental footprint, and more. Whether these inefficiencies are big or small, they all add up.
With 58% of companies planning on increasing their investments in AI this year, it’s essential these decision-makers know where their money is going. To secure a future where we use AI effectively and sustainably, we need to take stock of these issues now, before they balloon into insurmountable problems.
Computational Inefficiencies in AI
At the core of many AI systems lies a hidden waste of resources. Redundant calculations occur when models execute unnecessary operations during training and inference, often as a consequence of suboptimal algorithm design. Furthermore, there has been a prevailing trend towards increasingly large and complex architectures.
While a vast number of parameters might seem to promise better performance, over-parameterization frequently results in extra computation that yields minimal gains.
Along with these issues, inefficient data handling processes can lead to waste, like repeated data loading and suboptimal preprocessing. These factors lead to the ‘black box’ issue of deep learning, where the inefficiencies create needless complexities that make it difficult to identify inefficiencies, let alone address them.
AI’s Escalating Energy Drain
The energy demands of modern AI systems are a mounting concern that extends beyond merely high computational costs. Each AI query can involve multiple layers of intensive calculation, drawing significant amounts of electricity. The energy drain doesn’t stop at the training phase. The processing of the ever-increasing number of real-time applications consumes huge amounts of energy.
AI-specialized hardware already represents a considerable percentage of global energy consumption, driven by the increased popularity of generative AI, data centres, and crypto mining. That consumption is only going to increase as AI adoption increases, with the global energy use by AI projected to quadruple by 2030. Along with increasing AI’s carbon footprint from energy consumption, there will be increased demand for other resources, like rare minerals needed for hardware production and water resources needed for cooling.
The Economic Burden of Wasted Resources
... continue reading