The cost of AI

A variant on Moore’s law is the observation that the financial capital required to launch a new business decreases exponentially as technology gets cheaper.

Pre-internet, and using a notional geometric scale for the purposes of illustration, you might have needed $5m to found and build a new venture. The World Wide Web probably reduced that to $500k, while cloud computing brought it down to $50k. With the expansion of SaaS and API solutions, that cost might have been $5k to get going. Now, vibe coding and $500 of AI prompts can probably launch a new website, build a back end database, implement an e-commerce solution and deploy agentic AI bots to go and find your first customers.

This is a great outcome if measured by a lower barrier to market entry. It also enables founders to “fail fast, fail cheap”, and incentivises innovation by financially de-risking the process.

But even though the cost of AI tools is extraordinarily cheap in terms of the computing and processing power they deliver, there is a huge cost to our rapid adoption of AI that needs to be accounted for.

First, we are seeing corporate lay-offs among tech firms and parts of the service industry that no longer need as many human bodies and minds to operate at scale. So there is a human, economic and societal cost of increased un(der)employment.

Second, traditional skills and expertise are being hugely reduced in perceived value – why pay a graphic artist to design an image when I can use dall-e for free?

Third, as more and more creative tasks are being outsourced or delegated to AI (“create a short story about an F1 race in the style of Ernest Hemingway”) we risk losing our own innate creativity (that comes with experimentation, curiosity, play and reflection). This in turn devalues the creative process itself (thanks to cheaper, AI-enabled production).

Fourth, AI (and the Large Language Models on which it is trained) has no great respect for intellectual property. It doesn’t recognise boundaries between copyright material, content that is subject to creative commons, content that is in the public domain, and content which is publicly available. Again, if copyright owners and original content creators are not recognised or compensated for their work, why would anyone aspire to creating anything original?

Finally, there is the cost of resources (energy, water, rare earth metals) needed to maintain huge AI processing plants and data centres. (But at least this demand is accelerating the development of renewable energy.)

A few years ago, I posted a blog about the importance of the human factor, in the face of technological progress brought by automation and AI. I still remain cautiously optimistic that AI will bring huge benefits, despite the rampant growth of AI in the three years since I wrote that piece. But we are currently in an awkward and comfortable transition phase. If more jobs are lost to AI, and if human-led output is increasingly devalued, perhaps we will need to revisit the debate about Universal Basic Income and other policies to facilitate this transition.

Next week: Music, music everywhere…. and none of it very memorable 

Same old economic crises?

Amid the current turmoil surrounding tariffs and trade wars, I have been re-reading “Economics: The User’s Guide” by Ha-Joon Chang.

First published in 2014, this highly accessible introduction to economic theory and practice was written in the wake of the GFC, and the fallout that ensued from the US housing bubble and the consequential collapse (and public bailout) of major banks and financial institutions. The US bubble was largely caused by an imbalance in housing supply, poor lending standards, and over-engineering of mortgage-backed securities that quickly unraveled when banks lost confidence in each other, causing a major credit crunch and a lack of market liquidity.

Chang couldn’t have foreseen COVID and the knock-on effect on global supply chains and the impact of lock-down policies on overall productivity. He overlooks (ignores?) Bitcoin, a key ideological and technological response to the GFC, and he downplays the role of innovation in economic growth. However, his historical survey, his analysis of major economic theories (or “schools”) and his explanation of the roles that governments and the private sector play are all spot on and serve as a great resource for anyone wanting to try and make sense of the world.

Given the credit crunch at the heart of the GFC, the recent sell-off in the US bond markets reminds us that:

1. History repeats itself time and time again (albeit for different reasons)

2. Global markets are deeply interconnected, despite various attempts at de-coupling and policies designed to challenge globalisation and bring about increased protectionism

3. The US housing market is heavily reliant upon foreign investors since US treasuries both create market liquidity for new mortgage lending, and set key interest rates for borrowers – and major holders of US treasuries are foreign governments and institutional investors

The US mortgage market is underpinned by a near-socialist funding model (in the form of Fannie Mae and Freddie Mac), a propensity for long-term fixed rate loans, and a significant volume of non-recourse mortgages.

If a global trade war results in higher cost of goods for US consumers, and a bond sell-off results in higher interest rates, could we see a repeat of the GFC but driven by different causes?

Notes from the UK

I’ve just made my annual pilgrimage to the UK. It’s also 30 years since I emigrated, and with each passing year, I arrive feeling more and more like a visitor – although I am “from” there, I don’t always feel I am “of” there.

The following notes are some brief observations, in no particular order, based on a relatively short trip (2 weeks), and I was only in the Greater Manchester and Greater London areas.

  • I flew from Melbourne to Manchester, via Hong Kong. My in-bound flight to the UK was carrying a large number of overseas students from China – not surprising, as Manchester has one of the largest Chinese communities in Europe, and the city also boasts a UK Top Ten University.
  • A pint of cask ale in the north west cost me an average of GBP4.00 – in London, it was more like GBP6.00. I know some employees receive “London weighting” or a “London allowance” to cover the high cost of living, but I doubt salaries in the capital are 50% higher than the rest of the country. (Regional variations in property prices are a different matter altogether!)
  • On the other hand, a sour dough loaf from a local bakery in the Peak District cost me GBP3.00 – I would generally pay about 50-60% more for a similar product in Melbourne.
  • The in-coming Labour government, having won a huge majority in July’s General Election, has already hit the buffers. A combination of unpopular policies (cutting pensioners’ winter energy rebates), strange priorities (a ban on outdoor smoking), off-key messaging (“doom and gloom” rhetoric) and sleaze (donations of clothes, tickets and spectacles for the new Prime Minister and his wife) have brought the post-election honeymoon period to an abrupt end.
  • Staying with politics, there was a lot of despondency, if not anger, about the political climate. Despite Labour’s overwhelming success at the polls, it was hard to feel any love for the new government. And after more than four years since Brexit, no-one was jumping for joy at the outcomes, as the alleged promises and benefits fail to materialise. If anything, businesses are suffering due to the loss of access to EU markets and/or the additional costs of exporting.
  • Thankfully, the riots that erupted a few weeks ago have dissipated, but it felt like the underlying tensions remain. As well as having been triggered by malicious rumours and blatant disinformation, the social unrest revealed confusion about national identity (and what it means to be “British”), combined with contradictory views on immigration, multiculturalism and globalisation.
  • Meanwhile, the UK taste for “foreign” food continues unabated, along with a love of overseas holidays.
  • Despite producing some of the best television dramas in the world, UK content makers continue pumping out aging soap operas, stale game shows, endless talent contests and questionable reality TV. So, much like the rest of the world!
  • I paid GBP114.00 for a return train ticket from Manchester to London, which seems expensive for a 2.5 hour service. Both my outbound and return journeys were delayed by more than 15 minutes. Thanks to “Delay Repay”, I received a total of GBP42.00 in compensation. I can’t help thinking that the train and rail operators should focus on improving their services, rather than overcharging and delaying passengers, in the hope that the effort to claim is not worth customers’ time.
  • When visiting London, I usually use an Oyster card. This time, I forgot to take it – but thankfully, passengers can use contactless payment methods on trains, the Undergound and even short trips on buses (just remember to touch on and off with the same card on each journey!)
  • The autumn weather was especially mild, enabling me to indulge in long walks in the countryside, followed by a mandatory pint or two in a local pub (that great “British” institution!) Sadly, a combination of Covid lockdowns and changing social patterns means that many pubs have reduced their opening hours, or closed their doors for good.
  • As Australia’s near-duopolistic supermarkets face legal action for alleged misleading and deceptive price discounting, I’m reminded of the amount of choice UK shoppers have between supermarket chains, and across product ranges. No doubt that more competitive markets in Australia (for grocery shopping and beyond) would help alleviate the cost of living – but that requires structural and other changes for which successive Federal governments have had no appetite.

Next week: Does age matter?

Ticket scalpers? Blockchain could fix that!

Music fans of a certain age and demographic have been complaining loudly about the use of “dynamic pricing” when trying to buy tickets for their favourite band’s highly anticipated reunion tour. (There must be a pun in there about “Don’t book online in anger”?)

Part of the rationale given for using a demand-based pricing system is to disincentivise scalpers. The higher the cost of the ticket in the primary market (not the same as the ticket’s face value), the smaller the potential mark-up in the secondary market. Except that some tickets with a face value of $150 were priced at $450 at the box office, only to be re-advertised in the secondary market for several thousand dollars. In other words, the touts have simply increased their margins, in response to the so-called dynamic pricing mechanism.

Without offering any sort of apology or mea culpa, the said band have now announced additional tour dates, tickets for which will be allocated and sold in a form of ballot. Stop me if you think I’m being cynical, but by quickly adding dates to an existing tour itinerary, it shows that the band knew there would be excess demand, because it’s not that easy to reserve major (and highly profitable) venues, even 12 months in advance. And if they can run a ballot system now, why couldn’t they have done that in the first place?

All of which simply shows how out of touch bands like this are with technology and market dynamics. In short, ticket sales and allocations could have been achieved far more equitably if the band and their promoters had chosen to use blockchain, crypto and web3.0 solutions.

Here’s a simple list of options that could have been used:

1. Issue all tickets as NFTs (non-fungible tokens)

2. Limit the number of tickets per digital wallet and/or the number of wallets per ticket buyer

3. Ensure the use of soul-bound tokens to link wallet ownership and ID to specific individuals (to limit the number of tickets per wallet, and to limit the resale of tickets)

4. Run social media campaigns, quests and airdrops to allocate and distribute tokens that entitle holders to a place in the ticket queue – e.g., the more active a wallet holder is in the band’s fan community, the higher their chance of securing a priority place in the ticket queue

5. Pre-publish the expected ticket price ranges, and enable wallet holders to vote on the minimum/maximum price they would be willing to pay (using something like Snapshot)

6. Cap the amount an NFT-based ticket can be sold for in the secondary market or write the token smart contract to allocate a percentage of the resale value as a commission to the ticket issuer

Of course, the UK competition regulators are taking a close look at this ticketing fiasco, to see if so-called dynamic pricing breached fair trading or other consumer protection laws. If punters were not aware that they may have to pay far more than the advertised or face value of a ticket, this would appear to be unfair and unconscionable conduct. It’s potentially a form of under-quoting – advertise the ticket at a artificially low price, then force buyers to pay well over the face value at the actual point of sale (under the guise of “market demand”), knowing full well that the fans had little or no choice in the matter.

One final thought – knowing the volatile history of this band, the chances are that the concerts (or at least some of them) may be cancelled. Hopefully, the ticket agent and box office operators won’t be counting the advance ticket sales as recognised revenue, rather they are required to hold the funds in a verified escrow account until the performances are delivered and the ticket revenue actually earned….. (again, something that could be easily factored into a smart contract – no release of funds until the loud-mouth sings?).

Next week: Cooking the books?