Tipping points: looking at the sustainability of XR (metaverse)
I was fortunate to give a keynote along with luminaires like HE Eva Kaili, EU Vice President, Prof. Herman Russchenberg, and Pro-Vice Rector Magnificus for Climate Action, DU Teft, just to name a few. So honoured to attend this conference organized by the amazing John Havens, Maike Luiken, and Stefan Buijsman.
While there is tremendous opportunism on the metaverse, there are also some cautions if we are to enable this technology to unleash all its powers for good and the many (rather than the few).
A tipping point is defined as an inflection point, a point beyond which there is no return, but what you do now may make the difference between chaos and positive impact.
So some key points in my talk:
1. We are not there yet: It will take some time, as experts like Mathew Ball have pointed out – we are not there yet (more in our forthcoming new book). But this extra time is an opportunity to jointly think of the various scenarios where things can go wrong and hence build safety mechanisms. This thinking requires two things: First is brutal honesty (putting aside the hype). Second, we need co-creation. We cannot do this individually or even with a single organization, even if it is as powerful as a country or a block of countries. We have to do this together.
2. We don't have a common understanding of what the metaverse is: There are various opinions on what the metaverse is. Sometimes words are thrown around by investors and young startups – like Decentralised Autonomous Organisations (DAOs), Blockchain, and Cryptocurrency. All of which is not necessary for a metaverse. If you need to understand the metaverse, I suggest looking at science. In 1994, Paul Milgram and others came up with the XR (extended reality) continuum. At one extreme, you have the physical world or the real environment in which you and I are anchored. At the other end is immersive virtual reality (you can have 2D, 3D, and maybe 4D (if we think of time as different in virtual reality). In our earlier paper, Metaverse and Governance, my co-authors and I from IEEE looked at the metaverse as operating at various levels - parallel to, overlaid on, or interactive with the physical domain. If this is the case, metaverse is a part of the XR continuum. So it can use virtual reality and mixed reality and needs to seamlessly (in some form or the other) interact with the real world.
3. The tremendous speculation distracts from the work that needs to be done: This is simple to illustrate. Look at the media and market reactions to NFTs, the cybercurrency crash of 2022, and the crazy prices for virtual properties that have nothing to do with scarcity, like in the real world. The XR market is expected to be worth US$1 trillion by 2030, yet Citi values the market at 8-13 trillion by 2030. Considering our world GDP may hit US$100+ trillion at the end of the year, there is tremendous hope in how lucrative this new market would be.
4. Metaverse might exacerbate the inclusion problem: The metaverse needs high-speed, bandwidth internet and some significant investment in infrastructure and equipment. According to UNESCO, only 55 percent of households globally have an internet connection. In the poorest countries, this percentage drops to below 20 percent, and globally women are 23 percent less likely to have internet access than men. There is an inclusiveness problem when we look at the metaverse, even if we can use hand-held mobiles. The inclusiveness issues mentioned above do not address the data inclusion problem we already have and will have in the future!
5. Metaverse and human rights (like Article 23): One of the Universal Human Rights is the right to work (article 23). It is the underlying assumption of our education, economy, and wellbeing. But the metaverse will mean we need to rethink jobs. The growth in AI jobs in developed countries with an aging population and dwindling workforce may be at the cost of employment in countries with a youth bulge! Half of the global population growth between now and 2050 is expected to come from just nine countries! Though WEF says we will create more jobs than those lost, a recent study from the UK says that these new jobs are often less paying. Kate Crawford, in her book Atlas of AI, highlighted the extractive nature of AI and how it is usually built on the work of cheap labor through outsourcing or volunteered data and efforts. We make it easy as we produce 2.5 quintillion bytes of data every day (about 5+ million laptops)!
6. The beating heart of the metaverse is AI, so it suffers from all the challenges of AI: Here, I look at AI as a combination of hardware, software, and data (but, of course, data and software exploit human skills and human IP).
A. The hardware sustainability challenge: Sourcing resources for hardware and recycling is a vital problem. Our quest for economically mineable rare earth metals has taken us to space. We simply do not have enough rare earth metals needed for the exponential growth of XR. There is a noticeable shift in country strategies and policies from exploration to commercialisation and exploitation. And we have already begun littering space. The waste produced by electronic hardware (e-waste) is growing on earth. For example, even though 78 countries have legislation for e-waste, less than 20% is collected and recycled. Even worse, for example 60-90% of e-waste is illegally traded or dumped. The carbon footprint of our gadgets, the internet and the systems supporting them account may equal the airline industry and will double by 2025. The metaverse will increase our consumption of hardware (this includes the backend hardware connecting software systems, the IT and communications infrastructure, our gadgets and storage of data devices), not decrease it. Further the hardware challenge is compounded by trade. A lot of trade is cross-border trade (and often the parts criss-cross borders). This adds to the carbon footprint. The movement of intermediate goods, can account for up to 44% of world exports!
B. The software sustainability challenge: Mark Andreesen wrote an article in 2011, Why Software is Eating the World. Cost have come down, business models and government operations now revolve around software. It is estimated 70% of Fortune 500 use Microsoft 365 and 85% Microsoft Azure and 80% use SAP applications. During the pandemic – having a business online was in many countries the only way to survive. WhatsApp helped small businesses in India, and new payment gateways become more prominent. But there is a hidden cost to software, besides the issue of algorithmic transparency and bias. For example, each credit/debit card transaction leaves a carbon footprint, equivalent to an 8Watt bulb being left on for 1.5 hours. As we spend more time on our screens, software works behind the scenes and leaves an environmental impact. Cryptocurrency, gaming, watching videos, listening to streaming music….running a stockmarket….all have hidden emissions.
C. The data sustainability challenges: Much of the data we produce is unstructured (90%), which means that to make it usable, there is a significant amount of cleaning up and training or calculations. Running one large AI transformer model (213 parameters) may produce more than 5 times the carbon footprint of the life of a human. As we onboard new technologies that capture richer quality and greater volume of data, this will exponentially increase data requirements (capture, storage, processing and flow). Cross-border flows of data, are estimated to reach 65% of global GDP by end of 2022. The energy cost of data transfer and cloud storage is about 7-3 kWh per gigabyte (one hour of video) – which is a million more times the energy for your storing data in your hard drive.
7. Governance Challenges: Governance refers to the overall principles and processes that a country is built on, taking into consideration its political, historical, and social aspects to deliver public good [can be voluntary and legal] (From our latest book, p. 8). There are over 172 countries that use AI systems, and 60 have introduced 700 National AI strategies and policies according to OECD. However, only 42 countries have signed the OECD Principles of AI in 2019, and 25 governments (with EU) have joined the Global Partnership on AI initiative as of 2022. So there is a long way to go to get alignment across the world on good AI. The other important factor is the lack of time we have to work on challenges. Because the rate of technology adoption is scaling at an unprecedented rate (see the picture below - Pokemon Go took 19 days to reach 50 million users), we really need to work on transparently identifying and predicting unintended side effects and designing systems and processes for prevention, management and mitigation.
Recent Posts
See AllWe are seeing more movement of products across industries and the inability of regulators to catch these. Take for example the case of...
コメント