EV-Consumption Prediction :: Data :: Technology Impact Cycle Tool

Do you think this category is important? Not important A little A lot Very important

Depending on the technology you are assessing, some categories are more important than others. Here you can indicate how important you think this category is for this technology.

Skip this category? Yes No

If this category is not applicable for this technology, please select 'Yes.'

Are you familiar with the fundamental shortcomings and pitfalls of data and do you take this sufficiently into account in the technology?

Yes, I am aware of the pitfalls and understand that the data is never concrete and is always biased.
I took that into account by considering carefully only the parameters which  I needed and the ones that directly affect the charge in an EV. However, I still need to keep an eye on the model and it's performance during the production to catch any missing pitfalls.

(Only the gray shaded text is printed on the Quick Scan canvas)

There are fundamental issues with data. For example: - Data is always subjective; - Data collections are never complete; - Correlation and causation are tricky concepts; - Data collections are often biased; - Reality is way more complex than a million datapoints; Are you aware of these issues? How does the technology take these issues into account? We strongly recommend to do crash course four to properly understand the pitfalls and shortcomings of data (and it is fun!).

Go to the crash course:
Section Two - All data is subjective

(This question is part of the Quick Scan)

An example: Griefbot

Yes, dependent on the available data, the personality of the griefbot might be close or far from the deceased person. The limits are clear to us. The griefbot can't "cope" with it, so we would make the users aware of its limitations.

How does the technology organize continuous improvement when it comes to the use of data?

The technology works on the concept of retraining the model. 
And if the new model is better and more accurate than the previous one, the previous one is exchanged with the new one.

Data and insights about data change over time. If you measure something, chances are that you will influence the thing you measure. Does the technology takes this into account and how? Are you aware of the risk of a self-fullfilling prophecy? Do you organize feedback on the data? Is the feedbackloop closed? How does the technology take this into account? How does the technology improve on the way it handles data?

Go to the crash course:
Section Eight - Feedback loops

An example: Griefbot

We will let the griefbot develop itself, there will only be moderation on extreme new behaviour arisen from the self-learning of the bot.

How will the technology keep the insights that it identifies with data sustainable over time?

There is a possibility which suggests that the data will not be available in future, in case the clients do not agree to install the device in their vehicles which sends the data to us, in such a scenario the need for the application will also be over.

When using data to predict future behaviour, think about how sustainable these insights are: (1) what are the long term legal permissions to process the data? (2) can the data and used algorithms (if any) be kept current? (3) is the data available for a reasonable period of time in the future, even when the source generator(s) are discontinued or sold to a third party?

Go to the crash course:
Section Nine - Data is in the box

An example: Griefbot

The bot might develop itself to a future version of the deceased that might never have excisted. On the other hand it might not evolve, and keep certain believes in a time frame (eg. the view on same sex marriage), while if the deceased might have lived, they might have evolved in their believes.

In what way do you consider the fact that data is collected from the users?

The data stored by users is with their permission and they are aware of the data that is being stored. The stored data is treated with respect and is not for sale. However, it is possible that the insights from the data could be for sale.  But I do not see it happening at the moment.

If the technology profits from the use or sale of user data, do the users share in that profit? What options are considered for giving users the right to share profits on their own data? Do you think users are treated fairly? Why?

Go to the crash course:
Section Ten - The complexity of life

An example: Griefbot

There is no business model behind the grief bot.

Now that you have thought hard about the impact of data on this technology, what improvements would you like to make? List them below.

At the current moment, I cannot think of any substantial improvements, however, it would be a good idea to check the data before it is fed into the system, often broken sensors send bad data which is taken as correct data in the system since we never put a check and like to believe censor data is correct and will stay correct.
However, a broken sensor sending incorrect data could substantially affect the model and prediction system.
one thing I could suggest is to put a check in the system to ensure that the collected data is validated before it enters the system.

(Only the gray shaded text is printed on the Improvement Scan canvas)

If you think about the limitations of data. Things like subjectivity, incomplete datasets and so on. If you think about the way new insights are handled. If you think about the sustainability of the collection of data or the data that is collected from the users. If you think about all that, what would you (want to) improve? In the technology? In context? In use? The answers on questions 1-4 help you to gain insight into the potential impact of the use of data on this technology. The goal of these answers is not to provide you with a 'go' or ' no go' - decision. The goal is to make you think HOW you can improve the use of the technology. This can be by making changes to the technology or making changes to the context in which the technology is used, or making changes in the way the technology is used.

Go to the crash course:
Section Twelve - additional materials

(This question is part of the Improvement Scan)

An example: Griefbot

Not at the moment, no.

How do you rate the quality of your answers? Not good Can be better Good enough Very good

Are you satisfied with the quality of your answers? The depth? The level of insight? Did you have enough knowledge to give good answers? Give an honest assessment of the quality of your answers.

LTFE

Back

Frameworks

Ethical Data Assistant (DEDA)
(https://dataschool.nl/deda/)

5 Principles for Big Data Ethics
(https://towardsdatascience.com/5-principles-for-big-data-ethics-b5df1d105cd3)

Impact Assessment Mensenrechten en Algoritmes (IAMA)
(https://www.rijksoverheid.nl/documenten/rapporten/2021/02/25/impact-assessment-mensenrechten-en-algoritmes)

DATA GOVERNANCE CLINICS (DGC)
(https://research.tilburguniversity.edu/en/publications/data-governance-clinics-a-new-approach-to-public-interest-technol)

DATA PROTECTION IMPACT ASSESSMENT (DPIA)
(https://gdpr.eu/data-protection-impact-assessment-template/)

Datawijzer
(https://dataschool.nl/tools/datawijzer/)

Data Ethics Canvas
(https://theodi.org/article/the-data-ethics-canvas-2021/)

Articles

Big Data can't bring objectivity to a subjective world (from TechCrunch)
(https://techcrunch.com/2016/11/18/big-data-cant-bring-objectivity-to-a-subjective-world/)

What's Up With Big Data Ethics? (article from Forbes Magazine)
(https://www.forbes.com/sites/oreillymedia/2014/03/28/whats-up-with-big-data-ethics/#f8833fa35913)

Books

Weapons of Math Destruction (Cathy O'Neill)
(https://www.amazon.com/Weapons-Math-Destruction-Increases-Inequality/dp/0553418815)

Everybody Lies
(https://www.bol.com/nl/p/everybody-lies/9200000054655546/)