Long AI, Short Nature: Close the Short

While many are busy investing in artificial intelligence, the world is disinvesting in the intelligence found in nature. But the risk-reward of this ‘buy AI, sell nature’ is unattractive: now is time to close the shortⁱ. Here is how.

The metaverse is dead, long live generative AI?

It was only a year ago. Planet Tracker urged investors to forget the metaverse and instead invest in nature. Today, the metaverse has largely fallen out of favour and even though rumours of its death could be greatly exaggerated, it is definitely past the “peak of inflated expectations” and into the “trough of disillusionment” – see the Gartner Hype Cycle.

Figure 1: Recent press articles in English about the metaverse (click the picture to access the article)

This has been made more obvious by the contrasting, meteoric rise of generative artificial intelligence (AI) since 2022, made famous by the public release of OpenAI’s chatbot ChatGPT. Generative AI focuses on creating novel and distinct datasets such as images, videos, text and audio across multiple domains by using multimodal large language models (LLMs) trained on diverse inputs.

To understand how generative AI will impact our lives, Amara’s law is useful. It states that we tend to overestimate the impact of a new technology in the short run, but we underestimate it in the long run. Put differently, whilst we may now be at the peak of the hype when it comes to generative AI, we might not yet fully grasp how it will change our existence in decades to come.

Figure 2: Recent press articles in English about generative AI (click the picture to access the article)

And change our existence it will: since it is language that allows humans to configure and reshape society, an AI able to understand and use language can influence human emotions and behaviour (without needing to be sentient to do so). This can have extremely damaging consequences for humanity. Compared to historical technological novelties, the threat of AI is that it gives potential bad actors immense power that they have not had before. And, similar to dispelling fake news, mitigating its impacts takes orders of magnitude more effort in comparison to the effort required to create it.

This is why, according to a group of industry leaders including the CEO of OpenAI and Bill Gates, “mitigating the risk of extinction from AI should be a global priority alongside other societal-scale risks such as pandemics and nuclear war”. These risks include AI’s application in warfare (e.g. aerial combat, chemical weapons), or the spread of misinformation and power-seeking behaviour by AIs. They might sound like bad science-fiction, but they are coming fast to a data centre near you.

Yet according to a survey by BCG (Boston Consulting Group), workers on average are optimistic about generative AI, especially regular users as opposed to non-users, and leaders as opposed to frontline employees. 80% of the leaders surveyed say they use generative AI regularly. Only 20% of non-management employees say they do.

Generative AI is here to stay because we are wired to reward convenience

At the same time as they are warning us of looming threats, many AI industry leaders are seeking huge investments in this technology: OpenAI may try to raise as much as USD 100 billion in coming years, for instance.

Why invest hundreds of billions in a technology that could potentially destroy humanity, rather than shut it down, as many argue? As often, money is the answer.

Historically, the trajectory that every new technology follows is determined by its monetisation potential: ads for search engines, private data for social media, etc.

The monetisation potential of generative AI relies on our willingness to delegate tasks that require satisfactory completion. For instance, summarising long documents, creating designs and pictures, crafting social media posts, automating tasks, etc.

The generative AI market is predicted to reach USD 43 billion in 2023 and USD 98 billion by 2026. By 2030, AI could increase global GDP by up to 14%, as per PwC (although we understand that negative impact on GDP such as associated social and environmental ‘externalities’ have not been included). The bulk of this increase is linked to greater labour productivity. McKinsey estimates the latter at between +0.1% and +0.6% annually. Business functions expected to be impacted the most are sales, marketing, customer operations, software engineering, and product and R&D.

Essentially, monetising generative AI relies on identifying humans as expert in the use of the ‘satisficing’ strategy: aiming for a satisfactory or adequate result, rather than the optimal solution. People will use generative AI if the result is ‘good enough’, even if a human could have done a better job by spending more time. AI is convenient and new, so why not use it?

Recent history shows indeed that if something is convenient, people will be encouraged to use it: why else would we ride electric bikes and scooters, buy groceries online, talk to Amazon’s Alexa, drink coffee from single-serving pods or purchase pre-peeled oranges?

Convenient or essential? An essential question, an inconvenient answer

One major inconvenience of convenient products is that they often come with a burden of environmental, social or societal issues. This can also be the case for essential products and services (e.g. food or healthcare), but at comparable environmental or social damage, the damage caused by a convenient product feels worse than that of an essential one.

Of course, defining what is essential or convenient can be challenging since it is subjective and geography-, culture- and context-dependent. An access to shaded areas may be convenient in Iceland, but it is essential in much of Pakistan. Elevators are convenient for most, but essential when one’s mobility is challenged. During the Covid epidemic, the French government decided that cinemas, theatres and museums were not essentialⁱ, but places of worship were.

Something convenient today can become essential later on. Will generative AI remain convenient or become essential? Our guess is that its use will ultimately be so generalised in our professional and personal lives that it will be seen as essential by many. If you can remember a world without Internet, would you have guessed in the 1990s that it is often essential today?

However challenging, distinguishing between essential and convenient items is useful in that it allows for two clear realisations. The first is that it makes key metrics like GDP growth look oddly inadequate: how would economic growth look if it was stripped of convenient items? And as a consequence, how would valuation multiples change? Would Snap Inc (SNAP), for example, be valued at around USD 17 billion?

The second consequence is that it shows how essential things can be undervalued or not valued at all, and therefore underinvested. Examples sadly abound, but include: childcare, wellbeing, gender equality, air quality, etc.

The intelligence found in nature is essential but undervalued

Nature is another essential ‘thing’ that we undervalue. The current paradox of our fascination for artificial intelligence is that it makes our disinterest for natural intelligence even more acute. What is natural intelligence? Simply put, intelligence found in nature, where we define intelligence as the ability to acquire and apply knowledge and skills in order to fulfil goals. It is common in nature, at the species and multi-species levels:

• Raccoons are expert lock-pickers
• Bees can do basic arithmetic
• Octopuses are masters at escaping and solving puzzles
• Pigeons can distinguish a Monet from a Picasso
• Pigs can play video games
• Acacia ants attack herbivores that try to eat the leaves or bark of acacia trees, which provide food for the ants in the form of nectar secreted from the tips of its leaves.
• Within a few days, a brainless slime mould was able to re-create an interconnected network made of slimes that looks almost exactly like the efficient, well-designed Tokyo subway system.
• Microbes in our guts communicate with neurons in our brains that drive behavioural or emotional changes
• Plants can take controlled actions to mitigate diverse environmental stressors, discriminate positive and negative experiences and register memories from their past experiences.
• Intelligent, purposeful problem-solving behaviour is found in parts of all living things: cells, neurons, networks of neurons, viruses, RNA fragments, even in motor proteins and molecular networks.

Many of the above examples are purposely based on an anthropocentric definition of intelligence, as this is often how intelligence is perceived. It was indeed argued that this anthropocentric bias prevents the development of a general theory of intelligence capable of explaining the behaviour of not only human and artificial intelligence, but also any other entity that exhibits intelligent behaviour, such as nature.
Some cultures assign such importance to natural intelligence that they worship it. But in many cultures and in particular the Western ones, nature is instead ignored or destroyed. Why don’t we care more about nature in general, and natural intelligence in particular, even though our economic system is highly reliant on it? Maybe a comparison between natural and artificial intelligence will help.

Long AI, short nature: time to close the short

Below we show that whilst we invest in artificial intelligence, we disinvest in nature (and therefore in natural intelligence). But is this a good trade to make? We believe not.

Table 1: Comparing Artificial and Natural Intelligence. Source: Planet Tracker

	Artificial intelligence	Natural intelligence
Current value to society	Convenient	Essential
Availability to humankind	Typically privately owned, can be open-source	Typically freely available, but can be patented or privately owned
Production method	Invented and developed by humans at a high cost	Reproduces on its own in perpetuity if left undisturbed
Economic dependency	Low but rising fast	Very high
Potential contribution to future GDP growth	<2% increase on annual GDP growth1	Could be significant if Kunming-Montreal Global Biodiversity Framework goals are reached
Contribution to SDGs	Positive for 79% of SDGs, negative for 35% of SDGs	Positive
Contribution to employment	Neutral to negative	Positive
Annual public spending	Positive investment of tens of billions of dollars	Negative (destructive) spending of around USD 900 billion (c. USD 150 billion in spending offset by up to USD 1.1 trillion of harmful subsidies)
Private annual level of net investment	Hundreds of billions of dollars	Negative (up to USD 14 billion in investment offset by a much higher level of nature-negative investment)
Location of sources of capital	A list of 35 ‘plays’ on generative AI are all in the Global North	Globally
Explainability2	Low but growing	Medium but growing
Current phase of Amara’s law	Overestimation of short-term effect	Underestimation of long-term effect
Risk of enfeeblement for humans	Substantial	Non-existent3
Risk of destruction by human intelligence	Low but rising (via regulation)	Destruction ongoing
Extinction risk for humanity	Substantial	Non-existent (although humanity’s destruction of nature could lead to humanity’s extinction)

This is not to say that AI cannot bring essential benefits, including to support greater investment in nature, and nature conservation and restoration in particular.

But, when compared with artificial intelligence, the little importance that public and private capital give to nature in general and natural intelligence in particular is notably dire. Perhaps this is driven by the ubiquitous and free characteristics of all things natural? After all, the “problem” with nature is that it often solves its own problems for free. Instead, investing in problem-solving via technology offers a monetary return.

How can one invest in nature?

Nature appears like a tech company that initially made too many of its features freely available. It needs some “new” features which can be monetised.

Thankfully, there are plenty of ways to invest in nature and make a return. Below we have listed a few specific examples by main stages of the mitigation hierarchy, in addition to the obvious ones (reducing drivers of biodiversity loss such as pollution, land conversion and GhG emissions, and generally adopting more sustainable practices).

Table 2: The main stages of the Mitigation Hierarchy. Source: WWF/Planet Tracker

Primary goal	On land, you can invest in…	In the ocean, you can invest in…
Avoid impact	Textile and food traceability	Seafood traceability
	Alternatives to plastic	Recycling of shipping and fishing vessels
	Alternative proteins	Alternative proteins
Minimise impact	A deforestation-linked bond	Sustainable feeds for aquaculture
	A switch to continuous cover forestry	Monitoring of fishing fleets
	Water treatment	Noise reduction of shipping vessels
	Efficient use of water
Restore	Regenerative agriculture	Regenerative aquaculture
	Dam removals	More efficient marine protected areas
	Peatland restoration	Coral reef, seagrass and kelp restoration
		A blue bond to reverse overfishing

Finally, even without investing, there are two very powerful ways financial institutions could ‘invest’ in nature: by voting to align executive compensation with environmental sustainability targets, and by voting in favour of nature-positive proposals. Governments have a huge responsibility and opportunity to contribute too, by reforming the subsidies that negatively impact nature.

Decision-makers have shown they can act quickly, including reacting to the risks posed by AI. Planet Tracker therefore calls on decision-makers to invest in nature with the same urgency. It is time to stop biocrastinating. It is time to close the nature short.

ⁱ When trading assets, investors can adopt two main positions, a long and a short. The investor can either buy an asset – i.e. go long – or sell it – i.e. go short.