23 July marked my very first supervision meeting with Dr Hackl, and I want to preserve the memory here in a way that feels both reflective and a little raw. Instead of smoothing everything into a polished narrative, I decided to keep the long bullet points intact, as if they were stepping stones scattered along the path of this conversation. Each point is not just a record of feedback, but a small fragment of how I am beginning to see the project shift, refract, and open new directions.

  1. Dr Hackl was first curious about why I became interested in the topic of “evolutionary algorithms”. I answered that it was because I have long been interested in biology, especially evolutionary theory and its theoretical problems in the evolutionary process, and I had also studied it before. Later, in this year’s programme, I learnt various algorithms, and initially came to understand GA (Genetic Algorithm). I realised that although it is called “bio-inspired”, the evolutionary theory it draws upon is exactly the branch of evolutionary theory (the Darwinian paradigm) I had studied and found problematic. This made me start to worry: once certain scientific theories become “canonicalised”, it is very difficult for them to be questioned or challenged in the way that the humanities and social sciences are, since those are often regarded as subjective. Scientific theories, under the name of “objective science”, flow across disciplines, societies and technological applications, but at the same time diffuse their internal problems into our everyday life, society, and technological design imaginaries. This is precisely what Donna Haraway criticises about science as the modern “fetish”: because its seemingly “objective” knowledge characteristic serves as a cover, it is often accepted without being questioned. Even the general public often would not have the thought of trying to challenge scientific knowledge. I think that such a modern fetishistic worship of scientific knowledge, or its unthinking reception, also exists among contemporary AI developers. We often say that AI is very biased, as if it were evil, and that these developers are partial. But I sincerely believe that these developers are unlikely to be developing AI with the intention of leading humanity into a worse situation. That there are skewed results may perhaps partly be (of course there are many causes; I am only trying to address the one I am concerned with, namely the danger of the reception of “objectivity”) due to the uncritical acceptance of scientific theories and knowledge, which I see as a transformation of Hannah Arendt’s concept of the “banality of evil”: from her critique of the unthinking reception of morality and administrative orders, to the unthinking reception of seemingly objective scientific theories and knowledge. And this very banality may in fact be the root of many dangers (here I mean in the field of AI).

  1. Precisely understanding and using these complex algorithms and technologies is extremely challenging, especially on the practical level. The challenges lie roughly in two aspects: first, as I admitted to Dr Hackl in the meeting, many algorithmic data are not publicly accessible; second, even if one gains access to databases or files, they are often obscure and difficult, and there are not many people who can understand them in depth. For example, in the application of EA (Evolutionary Algorithm) in astronomical scheduling systems such as SPIKE, to what extent can I truly understand its functioning? Both “how to obtain the data” and “how to learn and master the theory and application of the algorithm” are urgent issues to resolve.

  1. My research is essentially a philosophical project, which gives me a certain degree of freedom: I can broaden the scope of cases, not being limited only to the SPIKE telescope scheduling algorithm, but also include other cases, putting them under the umbrella of my philosophical research question, and analyse them as different branches.

  1. Dr Hackl affirmed my simultaneous attention to “biological evolution” and “artificial intelligence algorithms”, and supported me in positioning the research as a philosophical inquiry. He pointed out that it is very interesting to think about these issues from a philosophical perspective: for example, the “natural selection” process in algorithms, and the concept of evolution based on the logic of competition – what differences and meanings do they have? This also means that my research does not have to be entirely confined to SPIKE, but can expand into wider philosophical discussions.

  1. If I were to use SPIKE as a case study, the prerequisite is that I must be able to thoroughly understand its functioning. But this is quite challenging for me, and it would be difficult to manage by self-study alone. Therefore, Dr Hackl suggested I consider conducting expert interviews, for example with computer scientists or other professionals in astronomy who use EA. The benefit of interviews is that they break down the barrier between theory and practice, make me clearer about the actual applications of the algorithm in telescopes, and help me overcome blind spots that are hard to cross with a humanities and social science background alone. Dr Hackl especially reminded me that although he could understand what I was trying to express in the proposal in general, readers might not be able to concretely understand the relation between EA and SPIKE scheduling, therefore I need to find more tangible ways to convey it.

  1. SPIKE is indeed an important case, but it need not be the only one. I should set my sight larger: take cosmology and astronomy as a whole, and my final research question to be framed, as the biggest umbrella, and under its branches explore more cases. In this way, my philosophical analysis will display richness and breadth.

  1. Regarding the format of the final project, I proposed in the draft to present it as “8,000 words written + comics as supplement”. Dr Hackl liked this idea very much, thinking that cosmology and astronomy topics themselves have many beautiful things to draw. Comics as a supplement could improve the problem of “insufficient tangibility” and “ineffective communication with readers” that usually accompanies the philosophical orientation of my work. But he reminded me: (1) if presented entirely as a comic, it would be very time-consuming, so the comic should only be a supplement to the written word, not a standalone piece; (2) this format idea needs to be confirmed soon with Ian (course director), because although he personally thinks it is good, Ian may have other opinions, and Ian’s judgement is very important as well.

  1. Dr Hackl particularly emphasised in the meeting that I must strive to frame my question. He said ideally, this question must reflect a puzzle that I am trying to solve as a specific problem. We read together the “Core Insight” paragraph in the proposal:

Algorithms are not neutral tools but embed specific worldviews. When we use evolutionary algorithms to schedule telescopes, we’re not just optimising efficiency. In fact, we are encoding assumptions about competition, value, and what deserves to be seen. This research makes these hidden choices visible and asks: what if we designed algorithms based on cooperation rather than competition?

On this, Dr Hackl pointed out: if I can provide a clear rationale explaining “why this matters”, that would be very good. And he believed that this “Core Insight” paragraph already is my final project’s core research question. He gave an example to help me frame it:

“How do evolutionary logics of competition apply in algorithms at the times of AI?”

Next, I need to think: under this core question, will I focus only on the astronomy project, or also consider other cases?

  1. Dr Hackl reminded me that the core research question (currently: “How do evolutionary logics of competition apply in algorithms at the times of AI?”) should function like a large umbrella. SPIKE is one case under it, but if practical difficulties arise, I should remain flexible, quickly find new cases, and not deviate from the research core. On later reflection, I realised this is why Dr Hackl put such emphasis on prioritising framing my research question during the meeting: because in this way, my final project has an anchoring point and will not drift, whether substituting new cases for old ones or running both in parallel – the core of the research will not produce inconsistency or incoherence.

  1. When choosing other cases, Dr Hackl stressed that accessibility is important, suggesting that I should prioritise cases where data and literature are abundant. For example, in the social sciences there are many studies related to AI, algorithms, large language models, labour platforms, gig economy, etc., which can provide easier points of entry. Literature in the social sciences is generally more sufficient.

Up to this point, most of our discussion circled around the telescope scheduling system, SPIKE, and the philosophical questions it raises about algorithms and evolutionary logic. But then the conversation shifted, due to Dr Hackl’s Eureka moment, almost unexpectedly, towards another field that touches me more personally: my own experience as an annotator. In that moment, the abstract puzzle of “competition versus cooperation” suddenly became entangled with lived human labour, with the messy and fragile realities of human–machine cooperation. This shift felt important, as if the project had opened a second door, one that could make my research not only broader but also more grounded and human.

  1. In the discussion with Dr Hackl, we found a new direction that especially aroused both our interests, and was also highly relevant to my part-time work experience: the role of human annotators in algorithm training. Dr Hackl further explained why he was interested in taking my work experience as the direction for a new case (this reasoning also made me reflect on how to make the final project more relevant. I think readers’ subjective suggestions—such as what makes something interesting, or what makes something feel relevant—are very valuable). He pointed out that, as an annotator, my past training and current experience precisely make “human–machine cooperation (especially data annotation)” into a field where the human–algorithm–AI relationship becomes particularly visible. This also echoes his earlier constant reminder to me: to use tangible cases to make the theory grounded, because for readers, visible and tangible material is more intriguing. Dr Hackl also shared his observations about annotation work: it is highly skilled labour, requiring annotators to handle complex tasks, and to possess high levels of concentration and judgement (for example bilingual ability—mine is English/Chinese—as well as multiple pre-employment language assessments and high-hour training). Ironically, under human–machine interaction, some basic annotation tasks are gradually being replaced by AI (because the machine is learning continuously). Dr Hackl then brought this observation back to my cooperation logic inspired by Margulis’s endosymbiosis: with the assistance of annotators, AI is indeed improving, and under annotator supervision is “evolving”. Therefore, he suggested that in the case selection of my final topic, besides the original discussion of “competition/cooperation” in the astronomical field, I should also add the dimension of “how algorithms evolve, and how, through human workers’ participation, their evolution is advanced—in a certain sense co-evolving with humans”. The further core question is: does this human–machine relationship resemble more the logic of “competition” or “cooperation”? Including these issues would make my case discussion more interesting and more resonant with the core research question.

  1. I also shared my two years’ work experience as an annotator: receiving the lowest pay, yet required to invest high levels of focus, skill, and judgement, and to be able to justify my annotations. However, there were often times of case shortage; at such times the company would provide a contract stating that if one wanted to receive more cases, one must “voluntarily” take annotation tasks involving pornographic and violent images. Some annotators, including me, under economic pressure, would make this trade-off and be forced to encounter content that could subsequently cause traumatisation. After the meeting, I thought of how the situation of highly skilled yet low-paid annotators resonated strongly with Donna Haraway’s discussion of the “women in the integrated circuit” in A Cyborg Manifesto: highly skilled women workers in the tech industry, especially those with high proficiency in English reading and writing, yet low-paid. Decades have passed, and the figure of the “women in the integrated circuit” has shifted from female workers in chip factories to AI annotators today, yet the structural situation of high skill but low pay has not greatly changed (this is also an aspect I can continue to dig into in my dissertation). In my dissertation, I can further draw a comparison between the labelling work carried out by scientists in the last century—as a form of self-affirmation of their esteemed trained judgement (Daston & Galison, 2007)—and the labelling work performed by the “women in the integrated circuit” in the era of AI.

  1. Dr Hackl responded: combining my experience with his concern for human–machine relations, he believed that contemporary human–machine cooperation (relations) are mostly unequal: annotators’ working conditions are harsh, wages low, quality and welfare inadequate, and occasionally the working content would trigger trauma (e.g. long-term exposure to pornographic or violent images). He linked these observations back to my research question, and argued that we could discuss: to what extent is the evolutionary logic of AI at the cost of human sacrifice? Hence, he proposed an interesting inspiration: to bring unions (I, as a non-native speaker, am not sure whether he meant a kind of union mechanism or labour unions) into the selection process and include it in my critical framework. For example, I could argue that human–machine cooperation in fact constitutes part of an “alternative logic”; if I can bring this perspective in, the final project would have more force, and be more interesting.

  1. At the same time, Dr Hackl seriously stated that current evidence shows that at least “for now”, such human–machine cooperation often still leads to inequality. Therefore, the relationship between “annotator–data–AI evolution” itself is difficult and complex. He threw this point out for me to consider further; after the meeting I continued thinking about it, and I believed that although the current human–machine relationship in annotation work is indeed, as he said, toxic and undesirable, this is precisely where the method of Utopia as Method (Levitas) I learnt in Utopia and the Future can come into play: attempting to construct an alternative relationship through the third mode, “architectural”. This method itself is also one of the main methodologies of my current proposal. Together with my personal working experience in annotation, this is why I have strong interest in this second new case direction Dr Hackl proposed. In terms of practice, another advantage of this case is that, as Dr Hackl said, it is more accessible, and so it also allows my philosophical ideas to be extended more fully. In sum, under the core research question, a case involving human actors makes the research more attractive and more relevant.

  1. Dr Hackl pointed out that even if, due to NDA, I cannot publicly discuss specific details of annotation work, this experience is still worth including at the “thinking” level. He particularly reminded me to pay attention to visual data annotation (for example, satellite imagery or other object detection tasks), and to ask how precise and accurate such visual annotation can be (e.g. detection of items of classified), and how this becomes part of the evolutionary logic of algorithms. If taken as a second case, it would not only fall under the same “umbrella of research question” as the first case, but also share the astronomy theme. This would allow me to further explore: in the value chain of human–knowledge production, what role do human actors play? How do humans participate and drive the evolution of algorithms? This value chain seems to be creating evolutionary logic—so what kind of evolutionary logic is this value chain co-creating? Can we allow this value chain to create another kind of evolutionary logic? And how can this be achieved?

  1. In terms of research design, Dr Hackl believed that the current proposal does not need major pivot. What is more important now is: I should first put forward a core research question, which should clearly point to a dilemma, a challenge, or, in EFI’s vocabulary, a kind of wicked problem; then design some sub-questions to unfold and deepen the discussion. Finally, Dr Hackl reminded me to seriously consider the delimitation of the research scope: should the final proposal put telescopes and astronomy cases into a broader comparative context? Or juxtapose them with one or two other cases? Or simply focus on the telescope itself? These orientations will affect the presentation of the research question and the overall framework of the dissertation, and are worth careful consideration.

  1. Regarding my advocacy in the final project of the cooperation/symbiosis logic inspired by Lynn Margulis’s theory of endosymbiosis as an alternative path in the evolutionary process, Dr Hackl suggested I read the novel The Swarm (a fiction about deep-sea life and cooperation between different species). I actually read it when I was a child, but at that time I knew nothing about the world and had no worldview; now looking back, I think this is a very good reading suggestion, and I will revisit it. Perhaps Dr Hackl only mentioned it because it came to mind, but after a year of EFI training I have adopted an ethos: to treat fiction (speculation) as a resource for imagining better futures or utopias, to treat imagination and speculation seriously. This also resonates with Donna Haraway’s emphasis on SF as worlding—especially alternative worlding. Over this year I have indeed experienced that many EFI courses include fiction in the scope of academic reading: for example, Utopia and the Future involved reading a large number of poems, and encouraged us to imagine utopias through poetry (the instructor said poetry is a forgotten magical ability); Coloniality of Data required us to read the novel Parable of the Sower and to learn poetic knowledge as a way of knowledge delivery; theIndigenous Futures’ instructor even actively argued that dreams should also be treated as serious data. Therefore, in terms of EFI’s ethos, I am very happy to revisit The Swarm, and, if appropriate, include it as material in the dissertation.

  1. Here is a speculation I made based on Dr Hackl’s feedback: the reason for showcasing two or more cases to present the logic of “competition vs symbiosis/cooperation” is that my original case of the SPIKE telescope scheduling mainly focuses on “pure algorithmic logic”. For most readers without a computer science background (including myself), even if they can imagine the basic concept of EA, it is difficult to truly grasp its concrete details, and so it is difficult to fully grasp my critique. Precisely for this reason, I need to introduce the case of human–machine cooperation (especially annotators boosting AI evolution): through this broader, more visible example, the difference between competition and symbiosis/cooperation among heterogeneous entities can be more clearly highlighted in the broader field of AI.

  1. Following from this, since algorithmic logic is difficult for both readers and me, Dr Hackl suggested that in the process of developing my project, I should contact some experts, and it would be even better if I could conduct interviews. He also suggested that perhaps I could send cold emails to authors who wrote about “how SPIKE uses EA” (in fact, I had already identified some in the proposal). He said he would be willing to look over my emails before I send them.

  1. I agreed that having conversations with experts is definitely a good idea. Such conversations can serve two functions: (1) as pure record, helping me to better understand the technology and theory; (2) depending on the case, as formal interview material included in the dissertation.

  1. Dr Hackl suggested that when contacting these experts, I should keep it simple and concise: clearly say that I am working on a philosophical project about evolutionary algorithms, and would like to arrange a conversation to better understand practical aspects.

  1. Regarding the missing blog post in the KIPP proposal appendix, Dr Hackl suggested that I could extend from the line of thought we discussed in the supervision, and consider what other cases related to the research question could be included. It does not necessarily have to be “EA telescope”, it could also be further afield, such as the aforementioned cases involving human actors.

  1. Before submitting the 2,000-word KIPP proposal, Dr Hackl is willing to look over my final draft.

  1. I also asked about the ethics form (which was unfamiliar to me). Dr Hackl responded: at present, my project seems to have no obvious moral concerns; since I have not yet decided whether to conduct interviews, the ethics form should be relatively simple at this stage. If in future I do need to conduct one or two meetings or formal interviews with experts, I can then amend it. Overall, the ethics form is not a major issue; once I have advanced some of the preliminary tasks of the final project, Dr Hackl can also provide further suggestions on the ethics form.

  1. Dr Hackl suggested that after finishing the final draft of the KIPP proposal, I could send it to him for review, and he said he looks forward to our continued communication and conversations. As long as time permits, he will try to provide help as needed.

  1. Conclusion | Immediate task list after the first personal supervision:

  (1) Complete the missing KIPP blog post (can absorb this meeting’s feedback and extensions).

  (2) Write to Ian as soon as possible to confirm whether the “comics as supplement to written text” format is feasible, and to seek his advice.

  (3) Contact professionals as potential interviewees (keep the letter short; Dr Hackl can look it over before sending).

  (4) Revise the 2,000-word KIPP proposal based on this meeting’s feedback (Dr Hackl can look it over again before submission).

  (5) Complete the ethics form (simple at present; amend later if interviews are needed).

  (6) Maintain continued contact and interaction with Dr Hackl.

As I read back over these notes, I realise this meeting was less about answers and more about learning how to carry questions. Dr Hackl’s guidance reminded me that what matters most is not the perfect case or the neatest method, but the ability to frame a problem that keeps pulsing with relevance, even as circumstances change. The telescope, the annotators, the algorithms—all of them are possible mirrors of the same puzzle: how evolutionary logics are imagined, lived, and perhaps reshaped. Leaving the meeting, I felt a quiet mixture of urgency and possibility, as though the research had just begun to breathe on its own. I look forward to seeing how the project will unfold in the coming month.

🩋 First Supervision Meeting with Dr Hackl – Feedback Overview and Reflections / Meng-Chieh Tsai | Cauldron of Dream Project Brewing by is licensed under a