As generative AI faces growing challenges, from unmet expectations to ethical concerns, some experts fear a new AI winter may be on the horizon. AI's rapid rise in recent years could be followed by a cooling period in 2024.

Are We Entering a New AI Winter?

Artificial intelligence (AI) has experienced several cycles of intense optimism followed by periods of disillusionment and reduced funding, often referred to as "AI winters." These periods have been characterised by inflated expectations of what AI could achieve, followed by a stark realisation of its limitations. Historically, AI winters have had profound impacts on research, development and public perception of the technology. However, with the rapid advancements in AI in recent years, it’s worth asking: are we currently facing another AI winter or are we on a different trajectory?

Early AI Winters

The concept of an AI winter first gained traction in the mid-1980s. However, the seeds of disappointment were sown much earlier. In the 1950s, AI pioneers like Alan Turing laid the groundwork for what we consider modern artificial intelligence, sparking excitement around the potential of machines that could "think”. The excitement culminated in the 1960s and 1970s with government-backed projects that sought to achieve human-level intelligence. Yet, by the late 1970s, it became apparent that the technology wasn't advancing at the pace that was expected. The systems developed during this time were far from the general AI envisioned. This led to the first significant reduction in AI funding and interest, marking the first AI winter.

The pattern repeated itself in the 1980s. Expert systems, which were designed to mimic human decision making in specific domains, were gaining traction. These systems, like Mycin in medical diagnosis, initially showed promise. However, the cost of maintaining these systems and their inability to generalise beyond narrow fields led to another collapse of interest and a subsequent second AI winter in the late 1980s and early 1990s.

The Resurgence

By the mid-2000s, AI was beginning to thaw. A key driver of this resurgence was the evolution of machine learning techniques, especially the rise of deep learning. Deep learning, a subset of machine learning, benefited from the confluence of vast amounts of data, powerful computing hardware and advances in neural networks. This combination led to significant breakthroughs, such as the development of convolutional neural networks (CNNs) that powered image recognition systems and recurrent neural networks (RNNs) that improved natural language processing.

One notable example of this renewed optimism came in 2012, when a deep learning system developed by Geoffrey Hinton and his team at the University of Toronto won the ImageNet competition by a substantial margin. This marked the beginning of AI's new "summer," a period of rapid advances in areas like computer vision, speech recognition and reinforcement learning, with applications from healthcare to autonomous vehicles.

Are We Heading Into an AI Winter?

Despite the astonishing progress in AI, the spectre of another AI winter looms over the industry. The remarkable advances seen in recent years, particularly in generative AI, may paradoxically set the stage for a potential slowdown. Several factors, including the gap between current AI capabilities and the expectations of (Artificial General Intelligence) AGI, as well as societal concerns, suggest that the AI field could experience a cooling period. However, the resilience of AI in today’s economy, coupled with its widespread adoption, suggests that a complete AI winter might be avoided. Instead, the industry could be entering a phase of recalibration.

In 2023, the AI landscape was defined by the explosive rise of generative AI, particularly models like OpenAI’s GPT-4 and its successors. These models showcased the power of machine learning in generating human-like text, creating art and even engaging in complex problem solving. This wave of generative AI brought with it a significant boost in public interest and commercial investment. Companies across various sectors—from entertainment to healthcare—began integrating these technologies into their operations. The promise of AI-generated content, personalised recommendations and automated customer service drove a surge in AI start-ups, with venture capital flowing into the sector at an unprecedented rate.

The rapid deployment of generative AI tools was also reflected in the market. According to PitchBook, venture funding for generative AI companies reached $1.37 billion in the first quarter of 2023 alone, highlighting the intensity of the AI boom. This period, often referred to as an AI summer, saw companies racing to adopt AI-driven solutions, hoping to harness the technology’s transformative potential.

However, as with previous AI summers, this boom was accompanied by growing concerns. The gap between the promises of AI and its current capabilities became more evident. While generative AI models like GPT-4 demonstrated impressive feats, they also had clear limitations. These systems could produce coherent text and mimic human conversation, but they often struggled with understanding context deeply or making reasoned judgments beyond their training data. This disconnect between expectations and reality raised questions about the feasibility of achieving AGI anytime soon. As a result, experts began cautioning against overhyping AI's capabilities, warning that unrealistic expectations could lead to another AI winter.

By 2024, signs of a slowdown began to emerge. While generative AI continued to make strides, the initial wave of enthusiasm had waned somewhat. Companies started to encounter challenges in integrating AI into their existing workflows and the economic climate became less favourable for the high levels of investment that had characterised the previous year. Regulatory pressures, particularly in Europe, also played a role in tempering the pace of AI development. The European Union’s AI Act, for example, sought to establish stringent rules for high-risk AI applications, adding a layer of complexity for companies looking to deploy AI solutions in the region.

Despite these challenges, there are reasons to believe that AI may not face a full-blown winter. Unlike in previous decades, AI is now deeply embedded in many aspects of the global economy. Its applications range from improving customer experiences on e-commerce platforms like Amazon to aiding in complex medical diagnoses. The healthcare sector, in particular, has been a major beneficiary of AI advances, with machine learning algorithms helping to identify diseases like cancer earlier and more accurately than ever before. Autonomous vehicles, powered by AI, continue to make progress, with companies like Tesla and Waymo pushing the boundaries of what is possible.

Moreover, AI technologies have become more accessible to the public. This widespread access to AI tools and education helps maintain momentum in AI development, even as the initial wave of enthusiasm begins to cool. Open-source AI models, such as those from Hugging Face, further contribute to the ongoing evolution of AI by enabling developers to build on existing technologies without the need for massive financial investments.

Join the IoA and prepare for the future of business

Sign up now to access the benefits of IoA Membership including 1400+ hours of training and resources to help you develop your data skills and knowledge. There are two ways to join: