Original rambling story telling here ...ChatGDP shortened and improved it in the previous post

AI-impact: 20-30% GDP growth a year or no more than 1-2% over 10 years?

Quite a difference between the Silicon Valley evangelists and the  pessimistic Daron Acemoglu as descibed in the Economist’s “Eureka all day long”. 

A very useful article indeed – that naturally has more questions than answers. To try to clear my thoughts I worked through my own Nordic experience, the reactions to the huge savings estimates along the line and how challenging it can be to take indirect dynimac impacts into account. And how those who would benefit most from modernisation often represent the strongest change resistance.  

This is especially important when looking at the full picture of the ID&AI phase. And maybe it can be useful also when Mr Acemoglu updates his estimate.

I. It started with e-banking in 1982 in Union Bank of Finland. We did not need to try to estimate the savings in those days. And the investments were small as core systems where already in real time and updating kept taking also e- and m-banking needs into account. 

Later the central bank estimated that e-banking had cut the cost of banking in Finland in half. Most of the cost savings went to customers via strong price competition. 

What should have been included in the Bank of Finland analysis – was (i) the time savings by customers who could now do their personal banking at home or at work - instead of travelling to branches (valued at 25€/hour in German impact analysis), (ii) the productivity improvements at work when time spent on routines could be used for productive work, (iii) the environmental impact from less travel, material and fewer bank branches, (iv) how internet usage benefitted from 20% of the grown-up population being enthusiastic e-banking users before Internet arrived (v) the value for society at large of the e-banking ecosystem and its user habit and trust as fundament for interconnecting customers with new services:  e-id, real-time e-commerce payments, e-billing, e-invoicing and e-salary. 

The sum of (i) economy of repetition for customers, (ii)  the economy of reuse of e-banking tools for new purposes, (iii) the economy of trust from banking, (iv) the economy of scope with a larger service menu, (v) economy of wider standardisation and even (vi) economy of scale became more evident. 

Something economists should have looked into (we did not have time)  – now also to have the perspective when looking into the coming ID & AI revolution.

2. The next hugh-savings-phase was e-invoicing – which started with customers complaining about having to take over the tedious job of branch and back office staff when filling in long reference numbers and other details in bill payment in e-banking. The solution was e-invoice-presentment with one-click approval.

 Then the State Treasury came out with a 150m€/pa cost-saving estimate for purchase invoices in state institutions. The municipalities arrived at the same 150m€/pa cost saving and then the Federation of Industries estimated that 2,8 billion/pa could be saved in enterprises. Over 3 billion € per annum in a small country.. (EU thus close to 250bn..).

The reaction by too many was to consider the estimates far too high – instead of seeing it as big enough for fast implementation.

What should have been added to these was (i) further routine work time savings by citizens, (ii) productivty improvements  when staff in all organisations could use more hours for productive work, (iii) cost savings also from sending e-invoices electronically, (iv) cost savings from less invoice-fraud and smaller grey economy, (v) the environmental impact from less material, transport and back office space, (vi) the value for society at large from e-invoicing becoming the fundament for automating accounting, VATreporting and furthering the Real Time Economy. 

3. The next phase started in 2006 was the public-private Real Time Economy program. Here the focus was to use e-invoicing and account statements as the fundament for full automation of accounting, value chains, VAT- and salaryreporting to a national income register. 

Each of these were estimated to have several hundred millions of cost savings from automation only. Implemention would have been faster if accounting firms would have been more active.

Again we should have included the impact of employees getting transferred from routines to productive work. Also better risk management and the value of data becoming available in - or close to - real time should have been calculated. 

A new opening for economic forecasting on both national and enterprise level  became visible with real time income reporting of salaries, pensions and  capital market income to the national income registry. https://www.finextra.com/blogposting/16304/four-big-screens

When this can be combined with real time automated VAT reporting anonymised data fundament can be made available to enterprises for better optimisation of procurement and marketing. 

It did become evident in our EU-work that the Single Market needs a decentralised ecosystem model and x-border functioning verifiable credentials to reach a European Real Time Economy Fundament for automation and new levels of data usage and security. 

IV. The next phase started with efforts to bring trust and privacy to the Internet with Self-Sovereign Identity and the Trust over IP model – a Trust Infrastructure (TI).  The public-private Findynet consortium started in 2018 aiming at implementing the MyData principles and making GDPR art 20 actionable with the organisation wallet applications. 

The importance of this was underlined by McKinsey estimating that the Trust Infrastructure could bring 3-6% GDP growth (depending on country) by 2030. 

The EU came out with only 2% possibly because EU started off with a too narrow wallet approach – basically only state issued identification credentials and drivers licences. Now it is being made clear that organisation/business wallets are the cornerstones (only way to get a critical mass of credentials flowing). 

What should have be included in the estimates are (i) the the time and stress savings in homes with general purpose trusted wallets  (ii) the importance of the TI for Cyber-security in face of cyber warfare and (iii) the explosive rise in cyber-fraud (now 11 trillion USD). The most important aspect is (iv) that safe and cost efficient deployment of AI-agentics can only be achieved with the Trust Infrastructure fundament’s  identity building organisation, citizen, thing and AI-agent wallets. 

Summing up how new fundaments have been created:

1. A decentralised payment infrastructure and standardisation tradition made e-banking possible in 1982 in Finland and was later taken to other Nordic countries by Nordea Bank.
2. E-banking became the fundament for interconnecting customers with e-ID, e-signatures, e-invoicing etc
3. This formed the fundament for the public-private Real Time Economy program – also adopted by the EU-commission
4. To make RTE functioning across borders a European/global the TI is needed as a fundament for exchanging data.
5. It has been widely understood that the TI fundament  is a precondition for safe and efficient AI-agentics

Taking the above experience of dynamic aspects into account I am sure that an impact analysis should deliver citizens-at-large  benefitting growth figures more than half way between the evangelists and no-growth scenario presented by Acamoglu. 

Especially as AI will not only free up time from routines but also bring much more power to ideation. And it will radically cut the costs of goods and services.