1. Digital farming
Farming underwent a massive transformation in the 20th century. At the beginning of the 20th century, farmers routinely made use of horses for their power, and often their own muscle in order to mow their fields. It was a labour-intensive discipline and very hard work. By the end of the century, farming made use of tractors, factory processes for livestock, and widely adopted pesticides.
The productivity improvement was huge. The absolute number of workers in agriculture had capitulated in the West, even though the population had grown plenty in that same period.
I envision a similar level of transformation happening again, where we might even see a near-fully automated and sustainable agricultural sector in 2030.
Internet of things technology is increasingly becoming more realistic, enabled by powerful chips (potentially made by Apple), ultra-long battery life (maybe designed with carbon dioxide technology), and super-fast network connection (maybe 7G wireless by this time). These can offer unparalleled analytics on the micro detail of anything, from a plant’s sensitivity to lighting conditions to its hydration levels during the day. It will make possible the ability to treat farms like scientific lab experiments, so that we can experiment and analyse all aspects of the crop over its life cycle. This information can be used to optimise the outcomes of crop growth and reduce our dependency on unsustainable technologies like pesticides, and also reduce our wastage of water.
Furthermore, the development of autonomous tractors, water sprayers, seed planters and tomato pickers will entail a huge swathe of automation, bringing productivity benefits while also making a large number of agricultural workers redundant. Combining the advanced analytics of farming with autonomous machines, we will see a large improvement in our sustainability of farming and also derive greater crop yields, while also helping us be more robust to changes in weather conditions.
We might even see the development of ‘skyscraper farms’, farms that are cultivated in large warehouses using artificial light and located closer to population centres, so that we can reduce transport costs, and begin to re-wild much of our planet and mitigate the catastrophic consequences of global climate change.
2. Blockchain networks
Today, a handful of companies have almost all of the data that society is generating. Much of this data seems useless: my time spent watching a DIY video on Sunday, or my location on Monday at 12pm. But amalgamate it all, and the predictive power of this data is immense. Their market capitalisations reflect this.
But this can cause problems. As Shoshana Zuboff writes in The Age of Surveillance Capitalism, market forces are pushing global tech firms like Google, along with smaller businesses trying to compete, into a race to surveil and intervene in as many aspects of our lives as possible, ideally without us knowing. The consequences of this are dire for democracy, with the possibility of the highest bidder becoming able to direct our attention towards different societal issues and blind us from others.
How do we respond? One answer is to make technology that enables digital users to own their data, and even be paid for its use. This will be cause for concern for current business models in Silicon Valley and elsewhere, where data collection is assumed to be nearly free and its use almost limitless. Hopefully it will encourage a radical rethink of tech firm business models, that incorporate societal goals for privacy, accountability and fair competition.
Blockchain networks might be the way we do this. They are built to avoid centralised control, are overwhelmingly expensive to hack and subvert, and can be designed in numerous different ways with different goals in mind. Their innate accountability opens the pathway for regulators to demand greater transparency from surveilling tech firms, and allow for society to have a more educated conversation on these issues.
One drawback at the present moment of blockchain networks is their computational demands. However, in 2030 I would predict that the computational capacity and energy efficiency achieved by gains in other areas will mitigate this issue enough to make these networks feasible and cheap.
3. Causal Artificial Intelligence
A human will only touch a hot stove once. By contrast, a neural network can make a dreadful mistake a vast number of times before it will change its behaviour. This is because neural networks need large amounts of training examples before they can recognise patterns. They do not perceive the cause and effect relationship that is obvious to a human in the hot stove example.
The necessity for large sets of training examples causes problems for computer scientists. Skilled machine learning researchers find their algorithms beaten in tasks by less sophisticated algorithms, simply because the latter has been trained on more data. GPT-3, Open AI’s natural language processing algorithm which has received widespread attention from the AI community for its ability to construct human-like written articles, was trained on the World Wide Web, the greatest artificially constructed data set in existence. Yet the researchers behind the project maintain that data is a bottleneck for the improvement of the algorithm.
While deep learning has revolutionised computer science and delivered outstanding improvements in pattern recognition, it is not going to be the model that delivers human-like intuition in the face of uncertainty (in situations where there are unknown unknowns). I believe that an AI model that can make causal inferences will be able to communicate better the mechanisms that led to its decisions, allowing researchers to enhance its capabilities in radically uncertain situations in, for instance, financial markets, automated driving, and public policy.
Research in this area has been heating up in the past year, and I expect that by 2030 this new breed of artificial intelligence will take bold new steps towards an AI capable of more than one task, or of being effective under radical uncertainty, where there are unknown unknowns.
4. Ubiquitous Artificial Intelligence
Now let us consider not just the heights of AI capabilities in advanced settings such as financial trading and public policy making, but also in general consumer application. In 2030, AI will power many of the services that humans do at the moment.
Chatbots will become the norm, so that when applying for an insurance coverage, ordering a pizza or getting legal advice, AI chatbots will be able to deliver many, if not all, of the needs you require, and seamlessly too. Self-driving cars will be safer, cheaper and more efficient than human-driven alternatives, and will be a normal form of transport around a city. Robots might assist disabled people in their day to day tasks, such as grocery shopping and going for a walk, and be seen travelling around a city with them.
AI will also be expert at many accounting, legal and consulting services, such that many professional firms will have to rethink their employment structures and expect different skills from their employees. AI will give you informed advice on how to trim the costs of your firm while abiding by tax laws, explain and summarise the legal components of a policy you want to pursue in your firm, or give you advice on business strategy using available methods.
In each case, the AI can explain the logical steps it took to arrive at these conclusions, and ask for feedback on the qualitative assumptions it used to build such hypotheses.
5. IoT Health
We have seen how the Apple Watch is able to save lives, because of its fall detection, heart rate monitoring and ability to make automated phone calls. There is scope for development in this space.
We might see development of nearly invisible wearable technology that can monitor our health data seamlessly, and provide users with informed analytics on how to improve their health. Doctors can make accurate diagnoses with more information from these wearables, and prescribe targeted drug treatments due to the individualised knowledge provided by this technology. Further, the need for surgical procedure may decline as the use of nano technology becomes sophisticated enough to handle procedures such as tumour removal.
This can mean less time spent on painful or uncomfortable treatments, greater effectiveness of preventative medical measures rather than reactive, and rapid discovery of cancer in the body.