Blog/eng

Covid-19 and Artificial Intelligence – a retrospective on the major trends

When Covid-19 started at the beginning of 2020, it was a bold and vibrant time for the AI community. On the one hand, it was clear that there is a growing need for automation in order to reduce physical contact and to enable a smooth transition into the work-from-anywhere mode. Facing these pressures, businesses were finally getting more accepting about deploying AI, while AI providers were continuously pushing the boundaries with new ideas. On the other hand, efforts in AI governance and regulation had been intensified in the preceding years. Thus, ethical and regulatory considerations imposed a certain restraint on the power of imagination of visionary AI entrepreneurs. All in all, you could feel a new creative energy despite the physical isolation of the first lockdowns. The industry as a whole flourished - as stated in the Tortoise Global AI Index, AI investments, which were already on a steady rise since 2010, more than doubled between 2020 and 2021.[1]

As our heads are cooling down about Covid-19 and the world is already tumbling into a new set of challenges, I would like to pause and take a look backwards. In this analysis, we will jump into a data-driven retrospective of prominent AI topics during the pandemic and see which insights we can take with us into the future. Note that, while the healthcare domain was certainly the largest beneficiary of Covid-related AI developments, the following analysis will focus on non-medical topics with a lasting impact for society and business.  

Identifying the top Covid-19 AI topics

The following chart shows the top-8 AI topics that have been particularly prominent in the Covid-19 context in terms of their share-of-voice. The topics are selected from a comprehensive classification of ca. 900 AI topics extracted from technology blogs and scientific literature since 2015. [2]

Figure 1: Top AI topics relevant to Covid-19 (outside the medical context)


Let’s consider these topics in the context of the pandemic:
1.    With school closures and lockdowns, the EdTech sector got an upwind. Beyond education for children and young people, many adults also invested the additional free time into learning, thus stimulating new technologies for continuous, life-long learning.
2.    Conversational AI and chatbots became popular in a variety of service domains incl. banking, healthcare and retail in order to reduce human contact and enforce social distancing.
3.    Data sharing came into play with contact tracking apps, the sharing of medical data on Covid-19 and certificate systems established in many countries. While indispensable for pandemic management and in line with the ongoing democratization of AI and data, this trend also raised many questions of data privacy and security that are now on the regulators’ agendas.
4.    Especially at the beginning of the pandemic, Covid-19 was a highly emotional, existential topic that dominated public media. As already noted by Chomsky[3], emotions lower our capacity for rational analysis and provide a fertile ground for spreading misinformation and fake news. Fake news detection was thus widely addressed in the AI literature.
5.    The negative economic impact of the pandemic forced both consumers and businesses to turn towards loans, thus spurring the development of more accurate and standardized credit scoring algorithms.
6.    Face recognition became more prominent and all the more controversial during the pandemic. Several countries including China, India and Poland used face recognition to ensure compliance to Covid-19 rules.[4] The rise of face recognition goes hand-in-hand with a general rise of interest in surveillance technology during the first months of the pandemic:

Figure 2: Share-of-voice of Surveillance over time


7.    Covid-19 came with an explosion in data quantity which can be matched up computationally with high-performance computing. The COVID-19 High Performance Computing Consortium assembles leading tech giants and universities in an effort to bundle global computing power for fighting the pandemic.
8.    The visionary concept of hyperautomation feeds the dream of the automated enterprise – and so far, it still remains in the future. On Gartner’s 2020 Hypecycle of I&O Automation Technologies, hyperautomation was well on its way climbing the peak of inflated expectations, while sliding into the “trough of disillusionment” already in 2021. This development is also confirmed by our data – we see a sharp increase in the second half of 2020 which, however, did not last into 2021.


Figure 3: Share-of-voice of Hyperautomation over time


Balancing out the AI revolution

Especially at the beginning, Covid-19 was a time of uncertainty paired with wild activity in the digital space. Facing this kind of intransparency, it is easy to loose control over the development and deployment of technology and allow it to gravitate towards unsustainable or even malicious use. Indeed, many of the topics pushed by the pandemic, such as face recognition, surveillance and the automation in healthcare, raised fundamental issues of safety and ethics. Concepts like hyperautomation and the growing substitution of service workers with robots and chatbots are disconcerting for our labor markets.
Fortunately, the past years have seen a surge of efforts in understanding the real-world implications of AI and ensuring that AI develops into a direction that is beneficial and sustainable for our society. In April 2021, the EU presented the Artificial Intelligence Act which lays out rules for the development, deployment and use of AI-driven products and services. Similar efforts have been ongoing in other major regions and markets. Over time, we might see an alignment of the national or regional policies and a convergence towards international standards.


Conclusion

Covid-19 has led to a wave of activity and investment in AI. In this article, we identified some of the hottest topics during the pandemic. These trends were driven by a range of new circumstances, such as the need for efficient pandemic management, the reduction of physical contact as well as living during an economic downturn. With joint developments such as data sharing, HPC consortia and common ethic standards, both technological and regulatory efforts might converge to a global AI agenda. Thus, while stricter border controls, the omnipresence of digital communication and disrupted supply chains are pushing us towards deglobalization in the physical world, Artificial Intelligence has the potential to become a truly global enterprise.

This article was originally published on The Yuan.

Notes and references

[1] Tortoise Media (2021). The Global AI Index, retrieved at https://www.tortoisemedia.com/2021/12/02/ai-boom-time/.
[2] The analysis presented here is based on the data and analytics behind Anacode’s AI Monitor. If you are interested in exploring the AI space using our monitor, please subscribe for test access to the monitor. If you are a data scientist, our Covid-19 Public Media Dataset (https://www.kaggle.com/datasets/jannalipenkova/covid19-public-media-dataset) gives you an opportunity to explore a large quantity of Covid-related documents on your own.
[3] Noam Chomsky (1997). Media Control: The Spectacular Achievements of Propaganda.
[4] Matthew Feeney (2022). Keep facial recognition away from Covid-19 response, retrieved at https://www.cato.org/blog/keep-facial-recognition-away-covid-19-response.