AI & The Post-Pandemic Future
I will be participating in a panel debate on this topic later this month, and I like preparing. As such, you can consider this article as opening up a train of thought preceding a discussion of the outlined title.
What is the post-pandemic future?
First it might be helpful to described what a pandemic is: “…an epidemic occurring worldwide, or over a very wide area, crossing international boundaries and usually affecting a large number of people”.
An epidemic is the rapid spread of disease to a large number of people in a given population within a short period of time.
In that manner this assumes that there will be a vaccine at some point and that it will get work, although it will take a long time and be a transition from a large amount of widespread cases to less and hopefully none.
After the pandemic and during
What I would say immediately is that there is another curve we should be worried about. This one is known to most of us as the climate crisis:
For EU one can see the conclusions from the special meeting
I would stress point A2 (added bold):
“The plan for European recovery will need massive public and private investment at European level to set the Union firmly on the path to a sustainable and resilient recovery, creating jobs and repairing the immediate damage caused by the COVID-19 pandemic whilst supporting the Union’s green and digital priorities. The MFF, reinforced by NGEU, will be the main European tool.”
In this sense both ‘green’ and ‘digital’ priorities are important.
EU announced 1.85 trillion euros with 750 billion for COVID recovery.
It is structured in the following way:
In this sense the two crises can be linked — using the crisis to assist in transitioning from damaging practices to more sustainable alternatives.
I chose to think of this as a ‘post-pandemic and during’ approach.
Resilient production networks and value chains
There is a discussion of making networks more resilient by Willy C. Shih in Harvard Business Review. He is the Robert and Jane Cizik Professor of Management Practice in Business Administration at Harvard Business School.
- New processing technologies.
- Continuous-flow manufacturing.
- Additive manufacturing (AKA 3D printing).
He talks about the dependency in supplier networks. One examples he gives is: “…production of the most advanced smartphone chips, which is concentrated in three facilities in Taiwan owned by the Taiwan Semiconductor Manufacturing Company.”
Further, I would say that digital value chains needs to be reconsidered.
What are these?
Servers, satellites, cables and so on — all these require a heavy amount of investment into sourcing and transporting resources as well as a great deal of maintenance. The operational cost of ‘data’ can be added to this. When I say cost I consider the cost to society and the environment to be important as a whole.
There are books surfacing increasingly that question or challenge the responsibilities of these large international value chains / supply chain.
World Wide Waste
Digital is physical. Digital is not green. Digital costs the Earth. Every time I download an email I contribute to…
Artificial intelligence in a post-pandemic future
Where does AI fit into this narrative?
I think in an overarching manner it will be hard to avoid within a variety of areas.
In relation to the increasing applications of AI and ML integrated into a variety of industries the computing needs in the overall digital infrastructure may shift slightly.
As such, the overall compute may change — as would be the case for the OpenAI and Microsoft supercomputer:
Microsoft announces new supercomputer, lays out vision for future AI work - The AI Blog
Microsoft has built one of the top five publicly disclosed supercomputers in the world, making new infrastructure…
Yes, we still need a great extent of nonvolatile memory (for storage of data), yet we may need more hardware that is custom built for processing data.
Particularly with new developments in compute such as the tensor processing units also to some extent known as AI accelerators. We may see larger facilities, and more of these, that are custom built to suit this purpose.
Highly advanced narrow AI will be more available in the coming years, such as is the case with GPT-3 from OpenAI that they are now openly offering on a commercial basis. Other large companies such as Amazon with voice and ownership of digital infrastructure; IBM with their large ML patent base; Google with their AI offerings and cloud; and emerging competitors within AI will be more commonly used by smaller companies.
Large submarine networks of cables are increase network capacity, also in areas often considered to have lower capacity overall:
2Africa is one of the largest subsea projects in the world, connecting 23 countries in Africa, the Middle East and…
Increased amounts of satellites seems likely as well, with Elon Musk’s Starlink being one example:
SpaceX is developing a low latency, broadband internet system to meet the needs of consumers across the globe. Enabled…
Jeff Bezoz has also mentioned that he wants to add a lot more satellites, with Amazon’s Project Kuiper — a mega constellation of 3,236 satellites:
Amazon Is Going To Add 3,000 More Satellites Into Earth's Orbit - And People Are Not Happy
The Federal Communications Commission (FCC) has approved an application by Seattle-based Amazon to launch more than…
What will it mean for people?
At this point it is easy to loose the outline of the inquiry. I mean, I am not a shaman or fortune teller. It is impossible for me to see clearly how this will turn out.
It does seem like there will be more Internet access around the globe. With increased connectivity and speeds AI and ML may gain even easier access as edge (distributed) computing with 5G and improved TinyML expands in the coming years.
Still, this will be highly unequally distributed. It is important to remember that just last year a little more than half of the world had access to Internet.
Look at 2019, and the ‘Users worldwide’:
One would think that with great spread of wi-fi comes great spread of freedom of expression. Not necessarily.
As mentioned by Vox in August 2020 the US banning of TikTok may spell more digital nationalism:
Within this narrative they mention the splinternet, the fact that Russia is also building its own separated infrastructure.
The 'splinternet' is already here
There is no question that the arrival of a fragmented and divided internet is now upon us. The "splinternet," where…
In that manner more personal control of digital influence seems likely.
Yet, with increasing availability of ML and AI it seems harder when state and private companies spend for influence.
Polarisation in the world and increased amount of authoritarian regimes is a trend that has been increasing as well.
A set of worrying trends for the post-pandemic future or the near present in the coming decade.
However, can we turn this trend?
Will we see a better world that takes sustainability into consideration to a much greater extent when digital technologies are applied, operated and built?
What will the post-pandemic future hold?
What do you think?
I hope you enjoyed this article.
This is #500daysofAI and you are reading article 460. I am writing one new article about or related to artificial intelligence every day for 500 days.