Enterprise Spending on GAI Set to Skyrocket to $143bn by 2027

Generative artificial intelligence (GAI) has been touted as the next big thing in technology, with leading figures such as Sam Altman, founder of OpenAI, suggesting it could revolutionise our world. However, while the hype around GAI continues to grow, it’s important to stay grounded and understand the current limitations of this technology.

Investment in GAI is on the rise, with projections suggesting that enterprise spending on GAI could soar to $143bn by 2027, a significant increase from this year’s $16bn. This surge in investment is fuelled by the promise of GAI and its potential applications. However, it’s crucial to consider the technological hurdles that currently limit GAI’s capabilities.

OpenAI, for instance, is actively seeking more funding to develop AI that can match human intelligence levels. This ambitious goal of creating a “superintelligence” – an intelligence that surpasses human cognitive abilities – is still a far-off dream. While AI models can make predictions, they lack comprehension. This fundamental limitation brings into question the feasibility of achieving human-like general intelligence with AI.

Take for instance AI text generators, which rely heavily on the data used to train them. These large language models (LLMs) perform well when dealing with recurring concepts but falter when faced with new scenarios or tasks beyond their training data. This limitation was evident when Google DeepMind’s AI weather forecasting model excelled at predicting recurring weather patterns but struggled to identify unusual or extreme events.

Furthermore, LLMs, including AI tools like AI text generator and AI images generator, have difficulty identifying their own errors. In a study conducted by Originality.AI, it was found that all LLMs produced errors. Even OpenAI’s ChatGPT-4 demonstrated inaccuracies in nearly one-third of its responses.

Despite these challenges, companies are exploring ways to leverage GAI for practical applications. These range from analysing employee performance reviews to scheduling waste collection times. However, the results have been mixed. A study by the National Bureau of Economic Research found that AI chatbot assistance improved productivity by 14 per cent, but these gains were mostly limited to new and low-skilled workers. Experienced workers saw little to no improvement.

As GAI continues to evolve and more AI tools are rolled out in 2024, these limitations will become increasingly apparent. This will inevitably place pressure on providers to address the critical issue of cost. McKinsey, a consultancy firm, suggests that AI could add more than $4tn to corporate profits. However, without clear pricing structures, companies are left in the dark about the potential financial benefits of AI.

In conclusion, while there is great excitement around GAI and its potential applications in the latest AI news, it’s important to remain realistic about its current limitations. As we move forward, it will be interesting to see how these challenges are addressed and how GAI continues to shape our world. For now, we can only watch, learn and adapt as this fascinating field of technology unfolds.