Are AI researchers independent?

10

October

2021

No ratings yet.

A few years ago a news article was published, which announced that the Netherlands should invest more into AI, Artificial Intelligence, research(1). In the upcoming year it is finally there and the amount of research into AI in the Netherlands will be doubled. The quality of the research has been high over the last couple of years. However, compared to neighbouring countries, such as Germany, England, and France, the Netherlands did very little in quantity. To change this situation, large investments are being made into the research of Artificial Intelligence. This research is largely paid by sponsors. The sponsors include companies such as Ahold Delhaize, Philips, DSM and Siemens. This poses the question if the researchers are allowed and able to conduct their research independently and thus without too much of an influence of the companies. Of course the sponsors do have a influence on the research, since it has to be relevant for them as well.

The concerns are not necessarily about the censoring of any outcomes during the publication. Everything is made public. But some researchers see potential problems in the different satisfaction with outcomes of a sponsor and a research. For example, if a company sponsor is satisfied with the results, but the researcher is not yet satisfied? How do you decide how to precede further.

Another example of this situation is given by Maarten de Rijke, who is the director of the national AI-centre. In addition, he is also professor Artificial Intelligence at the University of Amsterdam (UvA). He talks about algorithms that are being made to make recommendation to customers. For a company, it might be more than enough if the recommendations are being made and are correctly. However, a researchers might want to test if the algorithm also works under different circumstances or with different data. These problems can be solved by making clear agreements.

Overall, we can state that it is great to see that so much money is being invested into the research of AI. But researchers and business need to be aware of the dangers of such sponsorships. Even though there are many benefits, both parties may not have their interests aligned all the time. To still profit from these partnerships, clear agreements need to be made beforehand. In addition, we need to keep thinking about how to get the most out of these partnerships (2).

Sources:
https://fd.nl/ondernemen/1319074/nederland-moet-stevige-inhaalslag-in-kunstmatige-intelligentie-maken-ojj1cauO1Z3r
https://fd.nl/tech-en-innovatie/1415168/ai-onderzoekers-moeten-continu-hun-onafhankelijkheid-beschermen-njj1cauO1Z3r
Picture source: https://www.geeksforgeeks.org/top-5-trends-in-artificial-intelligence-that-may-dominate-2020s/

Please rate this

Data spaces. The Future of the Data Economy?

5

October

2021

No ratings yet.

Data is everywhere; companies are collecting a lot of data nowadays and it is getting more valuable. However, in most industries companies only have their own data. Usually, they don’t have access to the data of their competitors and vice versa. Of course, this is how it always has been and companies don’t just want to share their data with their competitors. However, this unwillingness to share, causes some unrealised potential.

Recently, a new type of company has arisen, whose mission is to make use of this lost potential. These companies are called data spaces. One of those companies, the International Data Spaces Association (ISDA), describes themselves as follows: ‘[We are] a secure, sovereign system of data sharing in which all participants can realize the full value of their data. International Data Spaces enables new ”smart services” and innovative business processes to work across companies and industries while ensuring that the self-determined control of data use (data sovereignty) remains in the hands of data providers.’ (1)

To summarize the statement above; data spaces create a platform to share data on for different companies. Part of what data spaces are based on is the development of AI, Artificial Intelligence. These algorithms need a lot of information. David Langley, professor digital innovation at Groningen University, describes it as follows; You need a lot of data to do something new with AI. If you combine different data spaces, you may come to insights that we can’t even imagine right now.

There is something very unique about the upcoming trend of data spaces, which is also described above. The user who shares its’ data has the full ownership of it and may decide who will get it and what to do with it. This is of course highly important, since businesses don’t want to share their company secrets.

This problem can also be described as a prisoner’s dilemma. Everybody stays in the situation which is the most unfortunate for all. Nobody wants to make the first move, since somebody else might profit from that, without giving back.

To solve this problem, there will be contracts and agreements to make sure that nobody can be taken advantaged of. Examples of the content of these contracts are the period of which a company may use the data, the encryption used for the data, how much is paid for the data and also how the data will be anonymised.

Even with the contracts, there is still a prisoners dilemma, but luckily there are more and more companies and institutions starting to share their data. For example, the Dutch intensive care departments have agreed to share their data to the ICU Data Initiative. The first results are already coming back. The different departments and approaches show that the amount of surgeries can go down after breaking bones. This can have a direct impact on the health care sector, because it shows the direct impact of different drastic surgeries.

The intensive care departments are however not commercialized, so I will also give an example of a company with the ambition to make profit, which is KLM, a Dutch airline company. Their goal is to predict for each individual component of an aircraft when it needs maintenance. The company already collects a lot of data; during a single flight 100.000 digital signals are measured in a Boeing 777-200. However, to reach their goal, they need a lot more data. Therefore, they have decided to join the AMdEX, the Amsterdam Data Exchange. They think that in the future everyone will profit from exchanging the data. Currently, there are no other airline companies who have announced to share their data, using data spaces, but as we said before; someone needs to make the first move. (2)

To conclude, data spaces are a very new concept and a lot still needs to be figured out. It can be scary for companies to make such a move and more detailed agreements and contracts need to be made to ensure the safety of the data. To see if it can actually have a long-term impact, we will have to wait until more results of the first trials come back!

Sources
1. https://fd.nl/tech-en-innovatie/1410279/waarom-klm-zelfs-data-uit-de-airco-verzamelt-wej1caZdkZa9
2. https://internationaldataspaces.org/

Please rate this