A unique feature of cutting-edge technologies like AI is their exciting and intimidating prospects. Recent developments in AI present both formidable threats and opportunities.The most outstanding characteristic of AI development is its exponential rate of advancement. Imagine a 700,000-fold increase recorded in the power of AI models in just over a decade. Recall how people were swept off their feet when, in 1965, Gordon Moore first predicted the computing power of chips to double every year. This contrasting rate of change represents the quintessence of what AI has in store. In the past five years, between 2019 and 2024, the capacity of large Language models to understand languages grew 10,000 times.Ray Kurzweil, Director of Engineering at Google, predicts that computers will have human levels of intelligence by 2029, as AI will pass a valid Turing test by then. A new technical term that has gained significance in the new millennium is ‘Singularity’. It represents a technological state when machine intelligence will surpass human intelligence.Kurzweil forecasts the singularity to occur by 2045. AI is not a single technology. The forward momentum in AI applications has been accelerated by a bundle of co-working technologies, from chip making to hyper-scale data centers with unprecedented data processing and transmission speeds.Initially, it was a struggle to develop learning models using statistical techniques to enhance the logical thinking abilities of the ICT systems. Then, it was the time for specialized chips that helped scale up the processing power of AI applications phenomenally.Between June 2019 and early 2025, AI-specific chip capabilities have multiplied phenomenally. For example, the performance of Graphics Processing Units (GPUs), AI’s nerve centers) has increased from 0.14 to 4.5 Petaflops, with their memory expanding from 32GB to 192GB and their memory bandwidth growing from 1 to 8 TB per second.This phenomenon is behind the sharp increase in the use of deep learning models in recent years.However, it is not the galloping speed of chips but the emerging synergies between hardware and software that greatly amplify AI’s future. A new breed of AI-based Electronic Design Automation (EDA) tools is helping enhance the capacity and performance of chips themselves. Simultaneously, as reported by MIT FutureTech lab, ‘AI could itself cause algorithmic progress’, creating a virtuous cycle.Even more significant is the surge in AI’s self-learning capabilities, which reinforce the reliability of autonomous or Agentic AI. Deep learning models distinguish themselves by their context awareness and ability to predict.Their self-supervised multimodal learning capacities also include reinforcement learning with human feedback. Progress in quantum computing and neuromorphic chips is poised to elevate AI’s self-learning to newer heights. We cannot, however, lose sight of the dark clouds on AI’s horizon. Large-scale loss of jobs seems imminent. AI is still in a nascent stage, but as it matures, it will also displace jobs in the service sector. Ethical concerns extend beyond privacy issues. Will it exacerbate inequalities in society? Will it abridge personal freedom by deciding on our behalf?By: Deepak Singh, principal solution architect, Gainwell Technologies
