There is little doubt that AI has the potential to transform the world economy. This has been reiterated in reports from the EU, IMF and major consulting companies and weekly in headlines around the world. As Lars Tvede has pointed out recently the perception is that: “(..)this is the calm before the storm. According to a recent McKinsey & Company study, “AI has the potential to deliver additional global economic activity of around $13 trillion by 2030, or about 16 percent higher cumulative GDP compared with today. This amounts to 1.2 percent additional GDP growth per year.”
But what if it’s just the calm before more calm. McKinsey said this before. Literally the same. In 2018 McKinsey estimated that GDP would be increased by 16% in 2030 due to the implementation of AI, while PwC at the same time put the target at 14% in 2030. McKinsey provided a helpful breakdown for how that would happen and interestingly an intermediate estimate for 2023, which provides us with an opportunity to test that assumption.
According to the estimate the increase in global GDP just by AI would be 1% in 2023 compared to 2018. If we look at official GDP data from the world bank we can get sense of that. The development in world GDP growth can be seen below:
The numbers show that over the period, AI has not made a dent in worldwide economic growth. Growth is the same or even a bit lower in 2023 compared to 2018. Surely if AI had been the transformational technology it would have been possible to see it by now.
if we look at the total growth of GDP we can get a better sense. In 2018 world GDP was $10,879 trn and in 2023 $11,567 trn, which amounts to a 6,5% increase in GDP over the period. If the prediction from McKinsey was correct about a sixth of all productivity gains should be attributed to artificial intelligence. That hardly seems likely in an era of unprecedented government spending through stimulus after Corona.
The optimism is intact and major investments continue, but it seems undeniable that there is an AI value gap, a gap between the expectations and the realities of AI. The economic value of AI simply has not materialized yet. There is no reason to doubt the economic possibilities of the smart people in institutions like the EU, IMF and the largest consulting companies in the world but there is reason to doubt that these will materialize automatically.
We are currently not on a path to harness the benefits that AI promises. 2 years after chatGPT there has been immense investments in AI, but AI has not even made a dent in the global economy since then. There is no evidence the situation now is different from 2018 when McKinsey and others made the same prediction. By now we should have seen an effect but since we haven’t. We need to look for reasons why that is not the case. As is the case for any technology, the value of AI is only realized when it is being used in the real world. But most AI projects gets stuck in beta. We should therefore look for possible causes for why that is. The issues holding back AI are not primarily technological but practical, human and implementation related. It seems like we are discussing the potential of the next series of powerful Formula One engines but no one is even focussed on assembling the car. That’s why we are going nowhere.
For almost 20 years I have helped organizations try to succeed with innovative AI technology. I have done this for public and private organizations, multinationals and startups and across diverse industries such as pharmaceuticals, financial services and retail. What is striking is that they all have the same issues: most of the good innovative ideas get stuck for seemingly different reasons. This is why we are not yet seeing any transformative changes to society or the economy. There is, however, a pattern to why innovations get stuck in the innovation pipeline. Understanding this pattern is the key to unlocking vast amounts of value by optimizing the innovation pipeline.
(BTW I wrote a short book about this last year)
English 
Danish