Beneficial AI, but AGI Worries

Beneficial AI, but AGI Worries 

An economist has to look at the wider informed opinion in a sector to understand how developments could impact economies and markets. Focused AI is already in use and provides efficiency gains in repetitive tasks. The 2020s is a period of rapid technological change facilitated by 5G rollout. This trend is fuelled by the interaction between AI and a surge in connection (Internet of Things), which allows a harnessing of data (Big Data).

This can provide economic benefits to countries that built the hardware and software required, or apply the technological improvements with a plan. The U.S. will likely be a net economic beneficiary from this theme, as it has been with other technology waves.

Focused AI can thus be net beneficial, but can cause economic problems if low and middle income workers skills are insufficient for the new jobs of the 2020s. Low and middle income workers could become unemployed and cause political tensions with technological improvement – Brexit and Trump are partially beneficiaries of a backlash against globalization and technology waves. This requires governments and large companies to have proactive training programs to reskill the workforce. However, DM governments have been distracted by COVID, then the Ukraine war/Energy transition and so reskilling the workforce is not a top priority. Focused AI development can still be a positive boost to the economy, it is just that it causes some economic side effects.

One topic of discussion among scientists and technology leaders is the potential misalignment of AGI. AGI is normally referred to as computer systems that are capable of generating new scientific knowledge and perform any human task. This is a high bar to achieve. However, previous estimates of a couple of decades are coming in as short as 5 years. In March, over 1000 scientists and industry leaders, including Elon Musk, called for a six month ban on AI more powerful than GPT4 (here). Worst case fears are that in 5 or 10 years that unless alignment with human values is forced on the industry that a powerful AGI could cause a major accident (e.g. AGI accidently removing oxygen from the atmosphere outlined in the 2021 UK Reith lecture (here).

The industry concerns over AGI reflect a lack of government regulation in the U.S.A split congress is unlikely to quickly build regulatory oversight, with the debt ceiling the current focus and then the 2024 presidential election. The EU is advancing regulatory oversight of AI with the European artificial intelligence act (here). However, Europe is a consumer of AI rather than the leading producer in AI. Finally, China is rushing to increase regulation for generative AI (here), which will include official review before generative AI is released. China authorities want to ensure that AI is aligned with the communist party’s objectives. It is thus the U.S. that has the major regulatory vacuum.

For economists and fund managers, the next 1-2 years will probably see the beneficial effects of focused AI coming through, but a watching brief is also required on how AI is regulated in major economies.