Professor from SGH deals with economic growth. Warns against the development of artificial intelligence

Prof. Dr. Hab. Jakub Growiec conducts advanced research on economic growth, particularly taking into account the role played by technological progress and human capital. The economist is also the author of the book “Accelerating Economic Growth: Lessons From 200,000 Years of Technological Progress and Human Development”, in which he deals with this issue on the scale of the entire history of the human species, indicating that we, as humanity, currently find ourselves in the digital era.

An important topic of his research is artificial intelligence and how its development can affect the world economy. Prof. Growiec talked about all this in the videocast “People of Science” produced as part of the series “Polish Science in the Footsteps of Copernicus”.

We invite you to listen to the entire videocast episode:

– I deal with modeling and explaining the sources of long-term economic growth – this is how the scientist spoke about his research in the video podcast “Ludzie Nauki”. – Speaking of economic growth, I also deal to a large extent with technological progress, that is, what causes us to have increasingly advanced technologies, which translate into increasingly higher productivity of economies, including economies on a global scale – he adds.

“Economic growth could accelerate to the pace of Moore’s Law”

Prof. Growiec emphasizes that artificial intelligence is becoming an increasingly important subject of his research. – Undoubtedly, because it is the most dynamic technological change that is taking place today and even if it does not directly affect GDP or the level of consumption, we can see that this technological change has great potential and can turn the entire economy upside down – he points out.

The economist is a supporter of the thesis that we, as humanity, currently live in the digital era. – What distinguishes the digital era, for example, from the industrial era, which was immediately preceding, is the use of machines to also replace or support mental work. While previously machines replaced or supported primarily physical work, it was in the digital era that mental work began to be supported – explains Prof. Growiec.

During the conversation, the SGH economist pointed out that although we do not yet generally see such a great impact of artificial intelligence on economic growth on a macro scale, “on a micro scale this effect can already be very dramatic”, and “automation of certain processes accelerates the operation of certain companies by leaps and bounds”. He cited the activity of Amazon as an example, which has forced a whole host of local bookstores out of the market.

– This does not have such a fundamental impact on a macro scale for now, because these branches are still relatively narrow. Also, artificial intelligence itself is sometimes very advanced for now, but usually in relatively narrow applications – the scientist lists. – Until these technologies are more general, and we are not yet willing to give them the power to make decisions, then growth will probably not accelerate so much – he adds.

However, the breakthrough point may be the so-called technological singularity, which can be understood as the emergence of artificial intelligence “smarter than humans, and in particular smarter than all of humanity combined.”

– This is the threshold where we expect that everything will really change fundamentally – indicates the professor from SGH. – And now, if we are talking about this technological singularity, it means in particular an acceleration of economic growth perhaps by an order of magnitude, to the pace of “Moore’s Law”, i.e. doubling of global GDP every 2-3 years – he indicates.

Prof. Growiec: We need time to solve the AI ​​alignment problem

However, apart from the profit that may be associated with such an increase, Prof. Growiec warns of negative consequences.

– Full automation will force us out of the job market, but it may kill us altogether. It may force us out of the world altogether. Because if we have a being smarter than us, which will take over all decision-making processes, then in the implementation of its goals our well-being will not necessarily be crucial – he warns. – Of course, what this intelligence will strive for will be a function of what will be programmed in it, and humans have control over that. However, humanity in general does not fully know what it is striving for and how to precisely define our well-being, especially in the long term. This problem is called AI alignment, or the problem of the alignment of artificial intelligence goals – explains the scientist.

That is why he places himself among the so-called “doomers” terrified by the development of artificial intelligence models and the race of individual corporations to obtain general artificial intelligence. However, in the conversation itself, Prof. Growiec indicates that it is still possible to limit the chance of a negative scenario.

– First of all, we need to control this disproportion that we see today, that the competences and capabilities of these algorithms are developing rapidly, while progress in the security of these technologies, in particular, I mean the degree of matching the goal function of these algorithms to the long-term well-being of humanity, is progressing very slowly – he points out. – Therefore, a postulate that is worth supporting is, supposedly a bit Luddite, supposedly a bit anti-development, but I believe that for humanity in the long term, a necessary postulate to inhibit this development of competences, to limit, for example, the computing power used to train the largest artificial intelligence models. Then, perhaps, this development of competences would be inhibited and we would have more time to solve the problem of AI alignment. This is a very key issue at the moment if we want to avoid these catastrophic scenarios – emphasized the scientist in the videocast “People of Science”.