Daphne Bennett

but now it is Tuesday

1. Tech skills

It will take a while to find the right pace. I had hoped to start on a Monday but I fell into a rabbit hole learning libraries in Python on VSCode. To overcome my psychological barriers, I decided yesterday morning was the first time I load some csv into VSCode and see what I can do about it with Panda and Matplotlib, I did some data cleaning and sorting etc. As my first selected dataset is a National Public Transport Access Nodes (NaPTAN), I discovered folium to create maps, then I discovered Open Route Service to get an API to create walking routes etc. Basically I messed around for the whole morning achieved nothing presentable. Back to work today and I have got a few monthly reports to submit and I hope I will have time to learn some new SQL syntax and familiarise myself with identifying info in their data warehouse. I am always too embarrassed how shallow I am – not just tech but in all other subjects, a mile wide, an inch deep. But the only way to change this is to be determined and invest time in them.

2. Masters studies – Ethical leadership

Who cares about ethics?????? I believe I get so touchy on this topic because I do care, too much, about ethics. Too much that it is difficult to say it out and discuss about it objectively. And from what I can see, no one give a shit.

I am still baffled by the way I was treated by the manager and the boss. I gained a bit more insight from Unit 2’s discussion about transformational and transactional leadership – and somehow it makes sense and reassures me I was not in the wrong.

We have started the 2nd Unit but I will try to stick to Unit 1. Unit 1 was mainly an introduction to the two notions of business ethics – dualistic and monistic, as well as the impact of globalisation and contradicting values across cultures. students are invited to think about what is ethics, whether we have come across unethical behaviours in real life and from our observation.

I did not want to dig too deep into personal life. I think about the news. I have written something about AI and Tech companies recently in the Durham University TechUP Programme, so this is a topic I have already done some research. This is my contribution to the discussion:

Last week, Meta was under pressure to explain its decision to quietly cancel a major contract with Sama, a third-party company it uses to train AI. The termination of contract will result in 1108 workers losing their jobs. It is not the first time Meta faces criticism concerning its ethical conducts related to data privacy and workers’ rights. How Tech companies operates internally and externally have long been controversial. I would say what data they use to train AI can be referred as internal organisational life, whereas how they outsource their data labelling and moderating supply chain globally is an external aspect of organisational operations.

Meta’s decision is made shortly after investigations published on Swedish newspapers that included the accounts of anonymous workers who were tasked to examine videos filmed by Meta’s smart glasses – including nudity and intimate encounters involving people who are unaware they were being filmed. These investigation and reports have prompted regulators to question how ethically the data is collected and how it will be used again. The UK data watchdog ICO had written to Meta to request information on compliance with UK data protection law related to AI-enabled smart glasses. AI-enabled smart glasses’ built-in feature can interpret images and sound, enable text translation and real-time Q&A about surroundings. When used wisely, it appears to be a life-changing technology, particularly for people who are blind or partially sighted. Users can activate recording manually or through a voice command. The issue here is that users may not be aware that their video and images could also be reviewed by humans, which Meta states it is outlined within its extensive privacy policies and terms of service. 

This decision also highlights the hidden external aspect of the labour systems powering generative AI race and concerns about worker treatment, primarily refugees and migrants, in the global AI supply chain. It is a common practice for large Tech companies like OpenAI and Google also subcontracts many workers around the world to label content manually, categorize photos and videos, and flag problematic content. Sama had previously worked with OpenAI to help filter toxic material for ChatGPT before its launch in 2022. Kenya-based workers were paid less than $2 an hour to review large amount of disturbing contents during their long shifts. Some workers reported PTSD symptoms from the assignments. Workers are offered little time for breaks and no rights to form unions. Any exercise of agency or resistance will cost the worker his/her job. Although it can be argued that data-labelling jobs provide new opportunities and stability in the developing world, help many individuals and their communities emerge from poverty,  the same inequalities had been played out in traditional outsourcing business like fashion and IT.

The lack of transparency of data and the exploitation of workers are only two examples of many moral controversies in the global AI industry ecosystem. The volume of sensitive real-world data collected by AI wearable devices, embedded in daily life that can continuously capture users’ surroundings from a first person perspective. This new technology appears to be a new frontier of the AI race and these problems will continue to intensify. A former worker at Sama said, “This issue is not confined to one company or contract. It shows how the global AI industry is shaped. Power sits with large technology companies. Risk flows downward, affecting outsourced workers, often in the global south, who have the least protection and highest exposure.”

Improvement of workers’ rights and data transparency are still necessary and urgent. As multiple players across the globe compete at breakneck speed, policymakers struggles to figure out how to hold these powerful companies responsible. It seems to be impossible to regulate the current AI companies with the structure of monistic mindset – that competition will take place under perfect framework conditions. Firstly, the technology is evolving too quickly for policymakers to understand and it often takes too long for actions to be taken. Secondly, many governments are highly dependent on these Tech companies for their day-to-day operation. Thirdly, it is difficult to regulate tech companies due to their international nature and their outsourcing practice enable cost-cutting and tax avoidance through legal loopholes. On the other hand, dualistic approaches – demand tech companies to consider both profit maximization and morality – appear even more unrealistic. Tech companies often argue if they follow the moral appeal, they will become vulnerable to their global competitors, including Chinese AI companies backed by CCP, therefore, gaining supremacy is a matter of survival. It is hard to see enough incentives for any of them not to exploit the immense power they have already held with billions of users and workers today.


Kenyan firm sacks more than 1,000 workers after losing Meta contract | Meta | The GuardianLinks to an external site.
Meta Quietly Ends AI Training Partnership with Sama After Claims Contractors Reviewed Intimate Smart-Glasses Footage – TekediaLinks to an external site.
Murgia, M (2024) Code Dependent – How AI Is Changing Our Lives. London: Picador

I was not quite certain if my response followed the guideline closely as it is more an observation than personal experience. But my bet paid off, my teacher said she has been teaching this module since 2022 and this is the first time AI has been mentioned in this discussion. It is a good example of globalisation and ability of large organisations able to outsource to countries with weak labour laws and low pay. It gives me some hope that ‘giving a shit’ matters. Later, I have also thought about other topics that I have gained understanding and experience through work, including Carers’ Allowance, PIP and water companies problems etc. This is a positive start of my path going back to uni. I will continue to make a habit of brain-dumping here.

3. Life and other things

When I am not rushing to work in the morning I decide I should walk the dog a bit further, not just come home as soon as she has done her business. I walk her when I drop my younger one at school. I think walking a bit more is good for both of us.

I have been drinking about half as I used to. I am trying to cut down screen.

Just me and the kids and remote learning remote working… It takes a lot of discipline to keep going. There could be an ounce of truth that Carl says I am one of the cleverest person he knows. Even if that’s not true, I still have to believe in it, I have no choice do I? I still couldn’t make out when Dylan said ‘ya we all agree you are a complete waste of space and in fact we are going to discuss how to get rid of you in the next meeting’ when I shared my vulnerability – I can also choose to believe that. Believing or not does not change the fact that I am still in my job and I have no choice but to figure out how to use these SQL syntax and the data warehouse system and how to link them to my PowerBI dashboards. I may try harder and work faster, or just keep doing what I am doing hopelessly – that’s my only choice. And in that sense, I don’t have a choice. I have to believe it is figure-out-able and I am clever enough to overcome my psychological and linguistic barriers to complete my master, progress in career and disciplined enough to look after the kids and myself well all that. Sometimes I let myself slip – but I have no choice but to pick it up again.

Life is easier when there’s no other choice. On that note, let’s get on with another day of work.