Some Thoughts on Bias

3 Replies Jonny Hankins

A Little story of bias

A father was driving his two children to watch a football match when they were involved in a terrible accident. The driver was killed immediately, as was one of the boys. The youngest child was sitting in the back on his car seat, survived the accident but was seriously injured.

The young child was taken to hospital where he was rushed into an operating theatre where they hoped they could save his life.

The doctor entered the room and looked at the patient, froze and said “I cannot operate on this boy, he is my son!’

Bias within Data

If you asked the question of how the boy could be the doctor’s son you are falling in a trap of bias. The doctor in the story is the child’s mother (obviously), but that may not be the first solution that comes to mind. In many societies we are brought up to see doctors as male, and nurses as female. This has really big implications if we are using computers to search for information though, as a search machine that uses content generated by humans will reproduce the bias that unintendedly sits within the content.

The source of the bias could be from how the system works. For example, if a company offers a face recognition service and uses photos posted on the internet (for example categorized in some way by GOOGLE), there will be a lot more white males than girls of Asian background. The results will be more accurate for the category with the largest presence in the database.

If a banking system takes the case of a couple who declare an income together, it will presume that the man’s income is higher then that of the woman’s and treat the individuals accordingly, because from experience the data shows that men’s income is higher then that of women and this generalization will become part of the structure.

The problem with language is also easy to see. If the example above of the doctor problem can be in some way ‘seen’ in the vast amount of text analyzed and used for an algorithm, then proposals and offers will differ according to gender.

Let’s take how we describe ourselves for a moment. A male manager will use a set of descriptive terms to describe himself that will differ from those used by a woman, he might be assertive, but she is more likely to be understanding and supportive. A system that unwittingly uses a dataset based upon (or even referring to) language used in job adverts and profiles of successful candidates will replicate a gender bias, because more proposals will be sent to people who use the language that reflects the current make-up of the employment situation.

In short: More men will be using the language that the system picks up on, because more men (than women) in powerful positions use that type of language. The bias will be recreated and reinforced.

In 2018 the State of New York proposed a law related to accountability within algorithms, Take a look at this short description, and the European Commission released a white paper on Artificial Intelligence – A European approach to excellence and trust in 2020. It might be more important an argument than it first appears.

There is lots of literature about this problem if you are interested, a quick online search will offer you plenty of food for thought.

Problems Counting People?

Leave a reply Jonny Hankins

Implementing COVID Regulations

Here in the Netherlands the University system just reopened after a short lockdown (again). There are still restrictions on how many people are allowed into rooms however, a maximum of 75 in any single space. This ruling was introduced last year, and led to some developments that might be of interest to technology fans (and privacy fans I should add).

Counting people manually as they enter and leave a room is a time consuming and expensive approach, so two universities had the idea of using cameras and artificial intelligence to check how many people are in buidings and individual spaces.

Utrecht University ran a trial, while Leiden placed 371 cameras on the walls above the doors to each space.

The Leiden approach however caused a bit of a stink. The cameras were all placed and set up while the students were locked out of the building, the ideal time we might say, to be having people up ladders in front of doors. But such an approach can also be seen as trying to do something without too many people noticing.

And that is how some of the students saw the arrival, and a couple started to investigate for an article in the weekly University student magazine.

Counting entries and exits, well nobody could be against that! The University has to do it by law. So discussion grew around the methods and the cameras and the data.

The university had bought 371 cameras from the Swiss manufacturer Xovis, 600 euro a piece. So the question is what can (and do) they register?

According to company spec, the system is capable of:

Counting students

Following their individual routes

Calculating an individual’s height

Estimating age

Suggesting mood (is an individual happy or angry)

Determining who is a staff member

Counting numbers in groups

Detecting incorrect facemask use.

Now these types of cameras are already in use in airports and shopping centres, to minimize waits (among other things) and to try and calibrate advertising and work out the actual moment that someone choses to buy something. So such data does offer broad analysis possibility.

The slogan used by the manufactures maybe lets the cat out of the bag a bit: ‘Way more than people counting.’

The cameras can of course be set to different levels of data collection and privacy, from level 3 fully anonymous (just numbers of people), to 0, which is a livefeed of the images.

Some Questions

Now I am no expert, but one problem seems to me to be that the system records lots of data, that at some point someone filters before providing their dataset to the customer. Who, when, under which circumstances, who manages security of access, there are a lot of issues here. But they are not all negative. Such a system may be of use in a terrorist incident for example, or other sorts of emergency. You could see why something more expansive might be chosen over a system that just counts movement. But there is a moral as well as practical dilemma in choosing such an overkill solution to a simple problem.

The report the student investigators published in the weekly university magazine showed lots of security issues, and there were protests from the students who wanted the system taken down. Both Utrecht and Leiden have now stopped using the cameras.

But that is not a good result from a responsible innovation perspective. Lots of money was wasted, many people got upset, two sides of an argument were constructed that are at loggerheads with each other.

A change in public participation techniques might have avoided all of this. A lesson to be learned I feel. Informing without debate doesn’t work.

You can read the student report here and a local newspaper report here. All in Dutch though, so you might have to use some translation software.

Is the pandemic over?

Leave a reply Christopher Roberts

England’s approach to COVID restrictions this January is very different to last January. It’s also worlds away from how European neighbours are reacting.

Many countries are now imposing tighter travel restrictions, and implementing lockdowns, while England (and to some degree the UK) is moving in the opposite direction.

For example, the “red list” of countries has been scrapped, as has the need to get a pre-departure test when travelling. Isolation periods have also been reduced and were masks not mandatory in indoor public spaces, you could be mistaken for thinking the pandemic was over.

England’s libertarian approach comes as the country’s infection rates hit an all time high. One in 15 people in the UK had COVID in the last week of December. Not since the pandemic started, or in the last year, in the last week! 🤯

So why is the UK making these decisions?

Do the statistics offer any justification for these changes?

Last year I posted several articles looking at the UK’s COVID-19 data and exploring the effectiveness of vaccination. Things have changed a lot since, so here’s an update.

UK COVID-19 Stats

The UK is now 90% vaccinated. Nine in 10 people aged 12 and over have had at least one dose of a COVID-19 vaccine. Around 80% are “fully vaccinated” having had two doses, and around 60% have also had a booster (or third) jab. 💉

While hospitalisations started to rise quite rapidly at the end of the December, they’re also still nowhere near to the 40k numbers we saw last January.

Why is this?

There are many reasons, but the two biggest seem to be: Vaccinations and Omicron.

Vaccinations

Vaccination has undoubtedly helped to weaken the link between infections and deaths. Despite there being around 3 million more infections in December 2021 than January 2021, there were around 30k fewer COVID related deaths.

This chart shows that link between cases and deaths.

December 2021 COVID-19 cases aligned to deaths

You can see last year in weeks 44 and 45 (January 2021) cases and deaths hit their peak – I’ve used this as the baseline maximum, 100%. Until week ~60 (May 2021) cases and deaths were fairly well aligned, cases went up, deaths went up. However that link has been slowly weakening. Since May 2021, cases have risen and fallen, with deaths hardly moving, and that’s in no small part thanks to vaccinations. By May 2021, around 1 in 3 people were fully vaccinated and 2 in 3 had had at least one dose.

In week 95 (the last full week I have data for) deaths were around 11% of January 2021 levels, while cases were almost 190%. The virus no longer has the same ability to kill as it once did.

N.B. Cases shown aren’t positive tests, but the ONS infection study estimates. Deaths are those within 28 days of a positive test, by date of death. Deaths have been moved forward by one week, to better align them to cases.

Omicron

The other contributing factor is Omicron. In the last month, UK COVID cases have been rising exceedingly fast. This is in part due to the more infectious Omicron strain of the virus.

5th of January 2022 COVID-19 variants by countryAt the start of December 2021, around 1% of UK cases were the Omicron variant, with Delta making up the vast majority of all cases. Last week, 96% of all cases were Omicron. That’s insane growth! Omicron took over as the dominant strain in around 2 weeks, almost wiping Delta infections out in the space of a month.

You can explore this more with this fantastic tool by Our World In Data – the University of Oxford.

Omicron appears to be easier to spread, more dominant, but less deadly. The levels of Omicron in the UK are surely also helping to keep deaths low – compared with if all cases were the Delta strain.

Do the statistics justify fewer restrictions?

So do the statistics give us confidence that England’s approach at the moment is well founded? To a degree, yes. It’s unclear if the decisions have been made based on science, or politics, but so far at least, England’s libertarian approach looks like it offers a good balance between freedom, autonomy and safety.

The more cases there are, the greater the risk of mutation. That could be seen as a concern, but mutation lead to Omicron defeating Delta, which (so far) hasn’t turned out to be a bad thing.

Is the COVID-19 pandemic over?

With more global cases than ever before, it’s undeniable that COVID-19 is still very much a pandemic. But, if we’re able to live with the virus in general circulation, without mass deaths or hospitalisations (just like we do with flu each winter) there is hope, that we may be nearing the beginning of the end of the period where COVID ruled our lives.

Live in hope. ☺️