blog

Automating societies: Nine predictions for 2020

By Nicolas Kayser-Bril • nkb@algorithmwatch.orgGPG Key

A few days before 2020, AlgorithmWatch makes nine predictions for the new year. Twelve months from now, we will come back to them and assess our forecasting skills.

In May 2020, Hawaii’s Mauna Loa observatory will record the highest carbon dioxide concentrations in the atmosphere since at least 2 million years ago. The observatory is one of the world’s primary benchmark sites for greenhouse gas monitoring. In May, carbon dioxide readings reach their yearly maximum. A new record has been set almost every year since measurements began in the 1960s.

Carbonated algorithms

The climate emergency led several organizations to focus on the link between carbon dioxide emissions and algorithms, which sometimes require gargantuan amounts of energy to create or operate.

AlgorithmWatch will begin a project on the topic next year, financed by the German Environment Ministry.

While it is unlikely that such projects will have much of an impact on greenhouse gases concentrations in the near term, the climate catastrophe will continue. In countries that have not upgraded their infrastructure to withstand the increased wind speeds, longer dry spells, heavier precipitations, larger fires and hotter temperatures, it is likely that communication and energy provision networks will fail. This will impact automated systems that rely on seamless connectivity such as automated border-control points or remote systems. Impact could be greater if large data centers were knocked out, as happened in 2012 when hurricane Sandy hit New York.

Prediction 1: Several organizations will run projects on AI and the environment.
Prediction 2: At least one automated system will be unavailable due to a climate-induced infrastructure failure.

Automated poorhouses

Despite the much lower hype around Artificial Intelligence, states and cities will continue to deploy automated systems in 2020, especially in the areas of welfare distribution, health and policing. Research by AlgorithmWatch showed that several police forces planned to introduce face recognition, which can be interfaced to other automated systems.

The British National Health Service recently signed a contract that offers Amazon access to patient data. One of the outcome of the partnership might be that Alexa, a voice assistant, offers automated health advice to Amazon clients.

Despite public outcries in several countries against such developments, few projects have been canceled. Automation is seen by many politicians as unavoidable and it is unlikely that the sentiment will change in 2020. Many of these developments will be featured in the update the UN Special Rapporteur on extreme poverty and human rights will probably publish, following his 2019 report.

Prediction 3: AlgorithmWatch will report on at least 50 new cases of automation in Europe.
Prediction 4: Fewer than 3 of these (or past projects AlgorithmWatch reported on) will be struck down as illegal by the courts or as politically unacceptable by local or national parliaments.
Prediction 5: The UN Special Rapporteur on extreme poverty and human rights will publish a new report that will again, and in colorful language, point to the role of automation in perpetuating poverty.

Platform opacity

The 2020 US election, on 3 November, might shift the political stance on automated decision-making. Several Democratic candidates have made critical comments on the issue. Elizabeth Warren slammed discriminatory algorithms when it was made public that Apple Card, a credit provider linked to the tech company and to Goldman Sachs, offered better conditions to males over females in at least two cases. Bernie Sanders called for a ban on face recognition in policing.

It is close to certain that campaign teams and other actors will use online platforms and their algorithms in an attempt to influence the vote. It is equally certain that we will not know for sure because Facebook, Google, Apple and Amazon will not grant researchers and journalists access to their data.

Prediction 6: Journalists, politicians, researchers and activists on all sides will claim that platforms and their algorithms influenced the 2020 election. Few of them will back up their claim with data.
Prediction 7: Before and after the election, platforms, especially Facebook, will claim to be transparent. They won’t be.

Regulation in 2021

In the European Union, public discourse in all 27 member states will continue to unduly focus on the situation in the US, including elections. But some organizations, AlgorithmWatch among them, will work hard at reporting on the situation in Europe. The “Automating Society” report, to be published in the summer of 2020, will be an important milestone.

It is unlikely that significant regulation on automated decision-making will be passed in 2020. The European Commission should publish a white paper on the topic in the first quarter, followed later by regulatory proposals. The Digital Services Act (DSA), which might include elements of platform regulation, will begin its long legislative journey at the beginning of the year.

Of the remaining large countries of the EU, none (bar Romania), are expected to have general elections. Sitting governments are either too feeble or too apathetic to carry out ambitious reforms on that front in the coming year.

While 2020 might feel like a legislative lull, it will let members of parliaments and civil society organizations produce strong and good policy proposals that might become law in 2021.

Prediction 8: The European Commission will publish a white paper on Artificial Intelligence in the first quarter, but no substantial legislation.
Prediction 9: AlgorithmWatch, among others, will publish a set of actionable policy proposals.

 


Photo by Charles on Unsplash (edited)
Published: December 30, 2019
Category: blog

Leave a Reply

Your email address will not be published. Required fields are marked *

AlgorithmWatch is supported by