Auteur: sneeuwwitje

Ik schreef dit korte essay in de zomer van 2024 voor Filosofie in actie blogpagina. De originele post is gemaakt op 31 juli 2024 en te vinden op de website van Filosofie in actie.

AI-solutions have been in use for multiple decades, but it wasn’t until OpenAI launched ChatGPT in late 2022 that the permeation of (generative) AI into our lives truly hit us. AI, short for Artificial Intelligence, is everywhere, from our search engines to our viewing recommendations and from algorithms that detect fraud to generative tools that re-shape the way we look at our creativity. In short, AI encompasses all types of “intelligence”. This increasing invasion of AI applications in society might turn more and more places into weird “other spaces”, or heterotopias, a space first described by Michel Foucault (1926-1984) during a lecture given to a group of architects in 1967. But what is a heterotopia and what does it have to do with AI?

The concept of the heterotopia is derived from the idea of the utopia, but unlike the utopia, which describes a non-existing (and ultimately “ideal”) place, the heterotopia is a space actually existing within our society, but outside it, somehow disturbing or upsetting what is outside of them. Foucault describes two types of heterotopias, namely the heterotopia of crisis and the heterotopia of deviation. Heterotopias of crisis are privileged, sacred or forbidden places, whilst heterotopias of deviation are spaces where we place that which deviates from our norms. Both, however, function as places containing that which would somehow upset or disturb the world without, such as menstruating women or the mentally ill.  

Even though heterotopias can be tangible places, such as cemeteries or boarding schools, one of the most striking examples of a heterotopia Foucault mentions is the mirror. In the mirror, you will find yourself and the space you occupy presenting itself to you – there, but not there. The space in the mirror is there, perceivable, but simultaneously this space is a virtual space, unreachable behind the shiny, smooth glass . 

In a sense, AI conjures heterotopias similar to the one we recognise in the mirror. The extreme use of algorithms in a desperate attempt to capture, calculate and conjecture every aspect of life constitutes a place imagined in the digital world, that we think accessible but is, in reality, as much in reach to us as the space within the mirror. It also contrasts with our world in that it is seemingly perfect and well-construed, promising a world wherein everything goes smooth and can be tweaked to fit our every desire and need. On top of that, Foucault describes heterotopias as places that are not freely accessible. Gaining access to them requires either coercion or the requirement to submit oneself to certain rites or purifications before gaining permission to enter. It is mainly the first condition that can be recognised in AI applications, as individual users seem to have little or no control over the ways their data is used to create the digital heterotopias dictating what we should watch, eat, or whether we are eligible for benefits or loans. 

The more we let our lives be dictated by these algorithms and applications, the closer we get to some absurd reversal of worlds. Instead of the heterotopia being the other space, the other space might become our analog world to which we “condemn” the people that are unable or unwilling to let their lives be modelled by a technology as slippery and cold as the mirrors in our bathrooms.

Photo credit: Michael Milverton via Unsplash

Dit essay werd in het voorjaar van 2024 geschreven door mijzelf en Joeri Hazelaar voor Filosofie in actie. De originele post is van 15 mei 2024.

Gespreksleiders, je komt ze waarschijnlijk steeds vaker tegen. Binnen allerlei organisaties laten mensen zich opleiden om een (ethisch) gesprek te kunnen begeleiden. Met deze ontwikkeling komen geleidelijk aan ook steeds meer vragen naar boven over het gespreksleiderschap zelf. Fundamentele vragen over welke rol een gespreksleider zichzelf in een gesprek moet aanmeten, of over welke kwaliteiten een gespreksleider zou moeten beschikken. Specifiek binnen het ethische gesprek komt daar nog een vraag bij, namelijk ‘Moet een gespreksleider over filosofische kennis beschikken, en zo ja, over welke dan?’

Vanuit ons perspectief als filosofen én gespreksleiders, zou het antwoord op de bovenstaande vraag een welluidend ‘Ja’ zijn. Voor het ethische (of ‘goede’) gesprek is een bepaalde mate van filosofische kennis wenselijk en misschien zelfs noodzakelijk. In dit stuk zullen we dit antwoord toelichten door middel van het uiteenzetten van drie typen kennis die relevant zijn voor het inhoudelijk vormgeven van het gesprek enerzijds, en het toelichten van de filosofische houding die van belang is voor het praktisch vormgeven anderzijds.

Drie vormen van kennis

Hoewel betekenisvolle en diepe gesprekken op allerlei momenten spontaan kunnen ontstaan, vraagt het ‘goede’ of het ethische gesprek een bepaalde mate van aandacht en voorbereiding: het is een weloverwogen gesprek, dat zorgvuldig gevoerd wordt, met een duidelijke intentie en een duidelijk doel. Er zijn drie vormen van kennis te onderscheiden die kunnen dienen als een theoretische basis die hieraan bijdraagt, te weten de conceptuele kennis van begrippen, de theoretische basiskennis van ethische perspectieven en kennis of bewustzijn van logica en retoriek. 

Laten we beginnen bij de conceptuele kennis van begrippen. Tijdens het ethische gesprek komen we op een zeker moment bijna altijd in aanraking met waarden, zoals rechtvaardigheid, autonomie of vrijheid. Waarden worden gedefinieerd als ‘nastrevenswaardige idealen’ en willen daardoor nog wel eens vrij abstract zijn. Om het gesprek voldoende betekenisvol te laten zijn en echt tot een gezamenlijk doel te komen, is het belangrijk om bepaalde conceptuele invullingen van deze abstracte begrippen te kennen. Zie hier de eerste vorm van kennis waarover de gespreksleider moet bezitten, namelijk de vaardigheid om deze abstracte begrippen uit de abstractie te trekken door het kunnen (aanmoedigen tot het) invullen van deze begrippen op verschillende manieren. Een mooi voorbeeld is het hierboven genoemde begrip ‘vrijheid’, dat in eerste instantie vaak een vrij duidelijk begrip lijkt te zijn. Toch zien we zowel in de praktijk áls in de filosofie dat er verschillende manieren zijn om dit begrip in te vullen. Isaiah Berlin laat dit in zijn bekende werk Twee opvattingen van vrijheid bijvoorbeeld goed zien: vrijheid kan volgens hem worden opgevat in negatieve zin (afwezigheid van dwang, hindering of beperking, ‘doen wat je wil’) of in positieve zin (waarbij men zichzelf verder ontplooit juíst doordat bepaalde kaders bestaan).[1] Door over deze conceptuele kennis te beschikken kan de filosofische gespreksleider herkennen dat het benoemen van abstracte waarden an sich niet voldoende is, en het gesprek op die manier verdiepen.

Om de tweede vorm van kennis te verduidelijken, zullen we twee vragen beantwoorden, te weten: 1) Wat verstaan we onder theoretische basiskennis van ethische perspectieven, en 2) wat voegt het hebben van deze basiskennis toe?

Wat betreft de eerste vraag kunnen we kort zijn. In de context van het ethische gesprek kunnen we zeggen dat de theoretische basiskennis op zijn minst enig begrip van de drie hoofdstromingen van de westerse ethiek beslaat, namelijk het Utilitarisme, de Plichtethiek en de Deugdethiek. In een breder scenario tellen we daar nog kennis van de zorg- of feministische ethiek en enige kennis van niet-westerse ethiek, zoals Ubuntu of Wu Wei bij op. Hierbij kent de gespreksleider op zijn minst de basale uitgangspunten van deze ethische stromingen en is deze in staat om argumenten die deze uitgangspunten volgen te herkennen.

Dit brengt ons naar de tweede vraag: wat voegt het hebben van deze basiskennis toe?

Vooraleerst valt te pleiten voor “begrip”. Kennis van verschillende ethische stromingen, draagt bij aan het herkennen van verschillende manieren waarop men ergens over kan denken, en hoe dit invloed heeft op de manier waarop men een vraagstuk aanvliegt of bijdraagt aan de besluitvorming rondom het vraagstuk. Het kunnen herkennen van deze gedachtepatronen maakt dat je mogelijk beter kunt begrijpen waaróm iemand iets zegt of een bepaalde keuze maakt, en je kunt daardoor een doordachtere reactie geven.

Een voorbeeld:

Begin februari van 2024 werd bekend dat supermarktketen Jumbo vanwege het hoge aantal winkeldiefstallen gaat experimenteren met AI om dit te verminderen. Hieruit zouden we een plicht-ethische blik op het fenomeen diefstal kunnen distilleren, namelijk dat het plegen van diefstal onwenselijk en slecht is – vanuit het oogpunt van bedrijven gezien zou je niet moeten willen dat mensen stelen, aangezien dat schade toebrengt aan het verdienmodel en tornt aan het bestaansrecht van winkels. Dit leidt direct naar een meer utilistische blik op de zaak; de utilist gaat uit van het vermeerderen van geluk en het verminderen van schade. Echter, deze schade kan vanuit twee kanten bekeken worden, namelijk zowel vanuit het perspectief van de winkelier die verlies maakt, als vanuit klanten die mogelijk schade ondervinden door het inzetten van technologie die vergaande controles uitvoert. Behalve privacy risico’s doet de inzet van AI mogelijk ook iets met bijvoorbeeld de autonomie van mensen; zij kunnen hun gedrag strikt aanpassen om maar binnen de kaders te blijven. Bovendien blijft de vraag wáárom mensen stelen met deze aanpak onbeantwoord, waardoor we helemaal niet weten of deze oplossing wel het beoogde effect zal sorteren.

Door kennis te hebben van de stromingen kunnen dus we beter begrijpen welke motieven de supermarktketen heeft om op een bepaalde manier te handelen, en tegelijkertijd geeft het ons handvatten om deze handeling kritisch te bevragen. Deze kennis zal de gespreksleider ten goede komen, doordat deze kan bijdragen aan het dichter bij elkaar brengen van deelnemers met verschillende perspectieven – bovendien draagt kennis bij aan het kunnen stellen van verdiepende vragen om het gesprek bij te kunnen sturen waar nodig.

De laatste vorm van kennis heeft te maken met het ‘technisch’ kunnen beoordelen van het gesprek. Kennis van logica en retoriek geeft de gespreksleider de mogelijkheid om argumenten goed te kunnen volgen en bovendien te kunnen beoordelen of de argumentatie juist verloopt, of dat er sprake is van drogredenen die het gesprek onwenselijk kunnen beïnvloeden. Als de gespreksleider deelnemers kan aanspreken op het gebruik van ‘retorische trucjes’ zal het gesprek van betere kwaliteit worden, omdat deelnemers gedwongen worden om goed na te denken over het geven van inhoudelijke argumenten.

Hiermee wordt – zo is de bedoeling – het goede gesprek gevoerd in plaats van dat er een overredingsstrijd plaatsvindt waarbij men hun gelijk willen halen. In het goede gesprek – waarover hieronder meer – staat het gezamenlijk vinden van antwoorden centraal. In een gesprek waar belangen en overtuigingskracht de overhand voeren en het gezamenlijk vinden van antwoorden naar de achtergrond verdwijnt, kan nooit bereikt worden wat een goed gesprek poogt te doen. Het is daarmee dan ook belangrijk dat een gespreksleider erop toeziet dat het goede gesprek gevoerd wordt en niet ontaard in debatwaardige taferelen. Echter, hoewel de theorie van redeneerkunde belangrijk is om te herkennen wanneer gebruikgemaakt wordt van retoriek in een gesprek, is dit nog niet alles. Naast de bovengenoemde soorten kennis is nóg iets nodig om het proces in goede banen te leiden, namelijk de juiste houding.

 

Een filosofische houding: gespreksvoering als handeling

Hoewel hierboven gerefereerd wordt naar theoretische kennis als basis voor een gesprek, speelt de filosofie misschien nog wel een grotere rol als het gaat om het aannemen van een bepaalde houding. Alvorens dit toe te lichten is een korte duiding van het begrip ‘gesprek’ in de context van dit stuk mogelijk nog nodig. Een gesprek kan verschillende vormen aannemen, maar wanneer wij spreken over het ‘goede’ of ethische gesprek doelen wij op het gesprek in de vorm van een dialoog, waarbij het intrinsiek onderliggend doel het gezamenlijk toewerken naar iets is. Het overtuigen van de ander of het halen van het eigen gelijk staat hierbij met klem niet centraal. Om het begrip van het goede of ethische gesprek en de rol van de filosofische houding hierin te verduidelijken, verwijzen we graag naar het zogenaamd communicatief handelingsbegrip[2] van de Duitse filosoof Jürgen Habermas. In het kort behelst dit in dat een goed gesprek een uitwisseling betreft waarbij uitspraken, beweringen, ideeën en argumenten bevraagd kunnen, en misschien zelfs moeten worden. Dit vraagt iets van de deelnemende partijen aan het gesprek, namelijk dat zij oprecht spreken en zich ontvankelijk opstellen voor vragen en kritiek, maar tegelijkertijd óók dat zij zelf kritisch zijn. We zouden dus kunnen zeggen dat om een goed gesprek te voeren, men zich een filosofische houding moet aanmeten. Deze houding omvat openheid (voor andere standpunten, maar ook voor het wijzigen van het eigen standpunt), nieuwsgierigheid en zorgvuldigheid.

De gespreksleider heeft daarin een voorbeeldrol; door open te staan voor alle argumenten en nieuwsgierig te zijn naar wat de deelnemers aan het gesprek drijft, kan de gespreksleider een veilig gespreksklimaat cultiveren. Uit deze houding volgt het principe van (zorgvuldige) reflectie. In combinatie met het kiezen van een gespreksmethode die aansluit bij het doel, vormt de houding van de gespreksleider de drijvende kracht achter het goede gesprek. Een mooi voorbeeld van het belang van deze filosofische houding voor het gesprek kunnen we terugzien in de overlevering rondom Socrates. Van deze ‘oerfilosoof’ werd gezegd dat hij zich na zijn terugkeer uit het leger dagelijks ophield op de Agora van Athene, en daar argeloze bezoekers aansprak met de meest uiteenlopende vragen. Socrates was echter zelden tevreden met het eerste antwoord dat zijn gesprekspartners gaven. Hij kreeg daardoor al snel de bijnaam ‘de Horzel’, omdat hij bleef doorvragen. Het gesprek, voor Socrates, ging over blootleggen van vooringenomenheden, het onderzoeken van aannames en ideeën en het analyseren van concepten, op een gestructureerde en gedisciplineerde manier. Het gesprek is daarbij niet louter gericht op de uitkomst, maar óók (of misschien zelfs juíst) op het proces. De rol die filosofie zodoende speelt, heeft minder te maken met het garanderen van een (‘juiste’) uitkomst, en meer met de manier waarop het gesprek wordt gevoerd om tot nieuwe inzichten te komen die uiteindelijk kunnen bijdragen aan een opvolgend doel, zoals het nemen van een besluit.

Conclusie 

In de vorige alinea’s zijn de filosofische houding en de typen kennis die je mee kunt brengen als gespreksleider besproken. Hierbij werd verondersteld dat het hebben van filosofische kennis van enig nut of belang is in het gesprek, maar is dit eigenlijk wel zo? Kan de gespreksleider, die toch vooral een faciliterende rol heeft, niet af met leiden op instinct?

Het antwoord hierop is genuanceerd: ja en nee. Je kunt best ver komen als je 1) al veel ervaring hebt of 2) een goede luisteraar bent. Toch kan de theoretische kennis een houvast bieden om bepaalde uitingen of argumenten te verklaren of beter te begrijpen. Het hebben van de juiste filosofische kennis kan bijdragen aan een goed verloop én een toegevoegde inhoudelijke waarde, doordat het uitnodigt tot het uitdiepen van concrete concepten in plaats van blijven hangen in abstracte ideeën, en uitdaagt tot het geven van (realistische) handelingsperspectieven in plaats van tevreden zijn met cliché-achtige uitkomsten. Aan de basis van het gesprek staat echter de filosofische houding – zelfs als we geen enkele filosofische kennis bezitten, kan het goede gesprek niet zonder een gespreksleider die deelnemers door middel van openheid, nieuwsgierigheid en zorgvuldigheid begeleid in een verkenning van hun eigen gedachten.

 

De vraag of gespreksleiders filosofische kennis nodig hebben stond in december van 2023 ook centraal tijdens een sessie op de professionaliseringsdag voor gespreksleiders[3]. Dit werd georganiseerd door het rijksbrede programma Dialoog en Ethiek, waarbij Filosofie in actie aanwezig was. 

Wij zijn ons bewust van het feit dat ons affirmerend antwoord op de vraag gekleurd is door onze achtergrond en ervaring; met het schrijven van dit stuk hopen wij de noodzaak en het belang van ethische kennis en het ethische gesprek toe te lichten en aan te scherpen vanuit dit perspectief.

[1] Zie Berlin, I., Twee opvattingen van vrijheid, Uitgeverij Boom 2010.

[2] Voor achtergrondinformatie, zie Habermas, J., The Theory of Communicative Action, Polity Press, 1989.

[3] Lees het verslag over deze dag hier: https://www.grenzeloossamenwerken.nl/actueel/nieuws/2023/12/21/gesprekken-begeleiden-is-een-vak

Ik schreef deze blog gedurende augustus 2023 voor Filosofie in actie, als vervolg op het ethisch onderzoek naar ChatGPT voor het Interprovinciaal Overleg dat ik samen met Piek Knijff en Joeri Hazelaar uitvoerde. Het origineel werd gepubliceerd op 5 september 2023 via Filosofie in actie.

Since the introduction of GPT-3.5 to the greater public by means of the ChatGPT chatbot, Large Language Models (LLM’s) have taken flight in people’s daily lives. As these things go, aside from ChatGPT, which is owned by OpenAI, other similar models popped up. Some of these, such as Bard and PaLM from Google or Claude from Anthropic are proprietary models like ChatGPT, which means they are owned by the company that made them, making technical specs like the source code, training data and Reinforcement Learning from Human Feedback (RLHF) data, walled off from outside scrutiny. Other models are open source models, some of which are trained on data from proprietary models made public, such as Meta’s LlAMa model, which made its specs public for researchers after a data-breach on 4Chan in March, resulting in the Stanford Alpaca model, amongst others. Hugging Face, a hosting platform, is currently the place to be to find any open source LLM out there. LLM leaderboards, which track, rank and evaluate LLM models on qualities such as performance benchmarks  and multitask language understanding can be found on Hugging Face as well as on LMSYS.org.

The open source developments on LLM’s are followed with curiosity, as open source software generally seems to have a more positive image than large corporate software shrouded in mystery. As seen in our July 2023 blogpost, Large Language Models (and ChatGPT in particular), cope with a broad range of ethical issues – ranging from controversial and untrustworthy output to privacy and copyright issues, to more “external” issues concerning climate and unfair treatment of workers. As LLM’s are well on their way to greatly impact the way in which we arrange society, using “responsible” models that mitigate risks of causing deeper inequality and unfair judgements to already disadvantaged groups are keenly anticipated by governments and educational institutions, amongst many.

This blogpost aims to shed some light on the meaning of open source in the field of Large Language Models, and to discover whether open source models are providing the “responsible AI” the world would like to see.

 

Definition of Open Source

Software distributed under an Open Source License is software that can be taken by users to use, study, change and distribute this software and its accompanying source code; it has little to no restrictions set upon the use and distribution of the software, as its goal is free access and all-round accessibility.

Open source software stems from the 1983 Free Software Movement, whose philosophy is based on guaranteeing that software users are able to use, study and share software freely, promoting collaborations between programmers and users. This movement was rebranded as the Open Source Initiative in 1998. The OSI is a public benefit corporation, trying to educate on open source and advocate for the benefits of having software that is open to everyone. On their website, an explanation of the definition of open source can be found, citing free distribution and (preferably free of charge) access to the source code as the first two criteria.

The accessibility of open source software provides certain advantages, among which are transparency, better security, flexibility and freedom from lock-in by companies. As the source code is open to anyone, oversight on security is higher, and security fixes and updates can be brought about by anyone, making software less susceptible to long-standing vulnerabilities of leaks due to inaction by companies. Moreover, due to source code being openly accessible, users can actually see for themselves how software “works”, and thus being able to verify if a product claim is true. This transparency might also give a hand in mitigating bias in software. As a result, software is easy to adapt to personal needs and public trust in software can be greatly improved – no luxury feat in a time where distrust in governments and companies due to intransparent use of algorithms and AI is rising. Open source Large Language Models may thus provide a valuable alternative to their closed off commercial counterparts, if they do turn out to be more open and transparent.

 

Open source Large Language Models (LLM’s)

As mentioned in the introduction, most open source models are fine-tuned on existing models. Fine-tuning LLM’s means training a smaller model on outputs of a larger model, thus improving the smaller model’s performances at a smaller cost than training a large model from scratch. Open source LLM’s like Vicuna (LMSYS), Alpaca (Stanford) and Dolly V2 (Databricks) are fine-tuned on larger models like LlAMa (Meta), a Pythia model (EleutherAI) and data from ShareGPT. Other open source models, such as BLOOM (BigScience), are pre-trained through so called Transformer architecture, similar to the GPT models from OpenAI.

Fine-tuned models appear to approach the capabilities of transformer and proprietary models such as ChatGPT, but still have some shortcomings (including most ethical aspects shared with proprietary models). In a May 2023 paper, one of the biggest shortcomings of fine-tuned models (also called ‘imitation models’) is said to be their poor performance when it comes to broad imitation. Smaller, fine-tuned models appeared to be approaching nearly the same accuracy as big models when it came to training them to perform a specific task, but quickly failed to improve on accuracy when it came to them performing a broad spectrum of tasks. According to the paper, to create a model performing well on a broad spectrum, a massive amount of training data was still what was needed. Underperformance seems to be confirmed on platforms such as Reddit, where developers have been struggling to get open source models work as good in practice as benchmarks claim they are. Another problem related to this is the fact that models trained in other languages can be underperforming, simply because for some languages there is not enough online data to be found to achieve the amount of training data necessary (a problem that Swedish researchers encountered, but managed to mitigate by using data from other North-Germanic languages such as Danish and Norse). However, considering the current rate of progress on LLM development, these problems might be tackled in the (very) near future, as also mentioned in a leaked Google document from earlier this year.

Another current downside of open source models is their User Interface (UI). ChatGPT and Bard have a user-friendly interface and provide API’s, which makes interacting with their models through a chatbot easy for anyone. Most open source models, however, don’t provide such User Interface, and require local installation and fiddling about with frameworks like LangChain to create an application. This, combined with underperformance, may greatly impact the implementation of smaller, open source models by the greater public who do not have the skills or means to figure out the specifics of this themselves.

Lastly, not all open source models are suited for every purpose. Models like Vicuna, Alpaca and BLOOM are all non-commercial models, mainly intended for research purposes. More on this topic can be found under the section Licensing and responsible use.

 

Openness is not always what it seems…

In July 2023, researchers from the Dutch Radboud University published a paper on the “openness” of LLM’s, concluding that models being presented as open source are not always as open as they claim to be. The researchers gathered information about openness on three different topics: Availability (openness of code, LLM and RLHF training data and weights), Documentation (licensing, code, architecture and preprint, available papers on the model, modelcards and datasheets) and Access (packages and an API for users). In their paper, they state the importance of openness when it comes to transparency, quality control and reproducibility of these LLM’s, all of which contribute to building trustworthy AI systems. In their research, it became clear that even though some smaller models provided good quality documentation, most open source models that were built on other, larger models would still be untransparent about their training data, simply because this data would come from closed off models to start with. Training data on human reinforcement learning also often remains unshared, and papers or peer reviews are absent for all models, with just some blog posts as reference material.

As long as LLM development is still largely dependent on the progress of larger, proprietary models, true open source might thus be difficult to achieve. Following therefrom, it is important to keep in mind that true openness should be interpreted as something broader than accessible source code alone, especially for AI software. Still, in an interview with De Volkskrant, one of the researchers pressed that using open source models with some shortcomings would be the better option when wanting to use LLM’s more responsibly.

 

Licensing and responsible use: the question of ethics

As mentioned above, a good amount of open source LLM’s are not fit for just any use, as they are created under the licensing conditions of the parent model. This means a lot of open source LLM’s are distributed under a non-commercial license and are exclusively allowed to be used for research and educational purposes, or personal use. In a June 2023 blogpost by the co-founder of GitLAB, AI licensing is explained to consist of 5 categories: proprietary, non-commercial NDA, non-commercial public, ethical, and open source. In the blogpost, it is argued that only licenses who meet the 10 criteria for open source software as defined by the Open Source Initiative (see above), are truly open. In this sense, Meta’s newly released LlAMa model released under a commercial license might seem open, but is still subject to restrictions and does not fulfill all of the OSI’s criteria for open software.

It is also interesting to pause on the distinct category of ethical licensing mentioned in the blog, which is described as follows: “The ethical license category applies to licenses that allow commercial use of the component but includes field of endeavor and/or behavioral use restrictions set by the licensor”. Ethical licensing is thus open in a sense, but closed in another, as the restrictions are based on ethical considerations and call for responsible use of the technology. Ethical licenses are grouped under the term Responsible AI License (RAIL).

Earlier this year, a paper was published on the adoption of Responsible AI Licenses for open source LLM models on the Hugging Face Platform. The researchers found the Open RAIL license to be the second most used type of licensing, even though they were uncertain if this amount is primarily caused by smaller models being finetuned on other models that are accessible under the RAIL license. Thus far, BigScience, the developer behind the BLOOM LLM and Stability AI seem to be the biggest players making their LLM’s available under an open RAIL license. On the RAIL website, several licensing documents can be found. Most of these seem to be focused on using the software in such a way that is does not violate legal rights and laws, and are formulated as such (see the End-User License, for example).

Another organization trying to foster ethical licensing, although not specifically focused on AI, is the Organisation for Ethical Source. They strive to center ethics in open technologies, and provide support in the development of ethical licensing with their “ethical licensure incubator”. Through this initiative, several licenses have been developed, among which are the Hippocratic License 3.0, the Do No Harm license and the At the Root license.

Initiatives like these provide hope that the way to more responsible LLM’s, and more responsible software development in general, will be breaking open. There is one sidenote to address, however.

The focus of the RAIL licenses appear to view “responsible AI” from a mostly legislatory perspective, while the Organisation for Ethical Source appears to focus mostly on Human Rights. Even though ethical issues can also be legislatory or Human Rights issues, to frame them purely within those bounds sells the breadth of ethics short. True ethics requires more than just asking yourself “is this legal?” or “do we comply with international agreements?”, but also requires both developers and users to ask themselves “is this desirable?”. Even if you comply with all (legal) requirements, there could still be reasons not to use LLM’s, for specific purposes, or even as a tool in itself. Using and licensing LLM’s in an ethical way would therefore require ongoing reflection and focus on topics such as (conflicting) values and the interests of different stakeholders. Subsequently, ethical licenses should strive to incorporate ethics in a broad way by and through encouraging this reflection.

All in all, even though for now the question on “responsible” licensing and responsible AI seems to be whether these are as ethical and responsible as they could or should be, the existence of open source LLM’s and initiatives concerning ethical licensing seems a promising and hopeful development. The future is ethical.

Disclaimer: this article was written during August 2023. Due to the rapid progress in the field of AI, some information may have changed as of publication.

Image: Maxim Berg via Unsplash.

Thoughts on access to culture for everyone

Imagine the following: I am a teacher and would like to watch a certain tv-series or read 

a Platonic dialogue to inspire my class and incite a good discussion between my students. However, not everyone (including myself) has access to the platform the series is streamed on, or school budgets do not allow me to purchase Plato’s book for all of my students. More than that, if I don’t touch upon this series or this novel, chances are that many of my students will never see or read them because they don’t know they exist, or don’t care and are not encouraged to try. My only option seems to download the series or get one copy of the novel and make a handout of the interesting chapters. However, if I do this, I am most definitely breaking the law…

In this essay I would like to explore if filesharing and downloading pirated versions of movies, music, e-books etc. should not be deemed illegal but rather seen as an act of beneficence to make access to arts and culture available for everyone.

Intellectual property

Intellectual property rights pertain to products of original thought, such as texts, art, music and so forth. The rights to these original thoughts protect the thinker or maker on different levels; they’ll be acknowledged as the creator and be protected from suffering (financial) loss because other people use their content without consent. In short, intellectual property rights protect the makers and thinkers from having other people unjustly profiting from their hard work. 

A (philosophical) argument justifying intellectual property rights can be found in the ethical branch of Utilitarianism.

Utilitarianism is a branch of ethics that states that what is good is that which delivers the maximum amount of happiness (“The greatest amount of good for the greatest number”). It is the result of an action that counts, and any action should strive to increase the happiness or the good for society overall. 

The Utilitarian-incentives-argument states that the protection of intellectual property is a necessary condition for ensuring that original thought is created at all. Why would people invest time and money into an idea, if afterwards anyone on the street can walk away with it and profit, leaving them with having done all the work but getting none of the rewards? Ensuring the creators get at least some compensation will, according to this argument, increase the number of awesome things and thus the amount of social utility.  

However, this leaves me wondering: would people with an artistic streak truly refrain from creating things and putting them out there for fear of shady companies printing their text or images on t-shirts, canvas bags and mugs, without them knowing and sharing in the profits? 

Definition of justice

I am an (aspiring) illustrator myself, and I create things and post them on Instagram and my online portfolio, not letting my lack of followers and the fact that I hardly make any money from my work stop me. I just want to do it. I like drawing, and I like making people happy with the stuff I make. Obviously, this does not mean that I would be happy to find my work being copied and sold without my knowledge, but the possibility of this happening is not stopping me doing what I love. 

I might try and protect my work adding a signature or a watermark on photo’s I post of any illustration I make, and some websites even prevent people from saving your photos to their computer. However, there will always be a way of someone being able to “unjustly” take your stuff and breach your intellectual property rights. 

What then, do we mean by “justice” in the cases of intellectual property?

In general, we define “just” or “unjust” in these cases mainly in terms of monetary profit or gain. Creating something other people might enjoy and having to distribute or share it for free while you’ve put your blood, sweat and tears in, seems to justify asking something in return for the fact you’re sharing. Especially in our current market-driven world, it seems you don’t get something for nothing, ever.

But this might seem like it’s just a capitalist argument made by shrew business (wo)men, while the intellectual property rights should plainly not be viewed solely in a money-making context (sometimes you make something to just cheer or help someone, not to get rich from it). So let’s define justice as follows: the maker or thinker created something that has the potential to get (widely) distributed, thus expecting a compensation or at least some recognition for their work or idea. The maker/thinker obviously has some individual needs that should be met. Thus, justice seems mean getting your deserts and needs. 

If we then view intellectual property rights as a form of justice, at first glance it seems to be a fair system. However, the laws defending intellectual property can only be called just if they truly and only serve the people whose original thought needs protecting (in any form). But what if we adapt the utilitarian argument, and look at the structures of the businesses who profit most from the rights nowadays?

The dark side of intellectual property rights

In recent years, it seems that the intellectual property rights no longer exist to benefit the actual maker or thinker, but rather the companies around those people. The most part of the money of Hollywood movies does not go to the people who created the idea or the script, but to the media conglomerates, just like the profits in the music industry flow to the record companies, with many artists complaining they don’t even own the music they themselves wrote and performed (two examples are Ke$ha and Taylor Swift; Swift even started to re-record her old albums so as to actually own the rights herself).

In these and other similar examples, it appears that the intellectual property rights do not concern protecting the actual original thought, but rather serve to make certain parties the most money possible. So in these cases, it’s doubtful whether the intellectual property rights called upon when it comes to downloading pirated movies, songs etc. actually are just.

Cultural capital 

Everyone probably has an anecdote from their youth concerning a fad you just had to be in on. Whether it was owning a Furby or Tamagotchi, or watching the most popular show on television to be able to join in on the conversation in the schoolyard.

Being part of something or having access to something is important for us on different levels. Pierre Bourdieu, a French sociologist, introduced us to the notions of social, cultural, and economic capital. The cultural capital is the basis on which the other two are largely dependent; the cultural capital you are given in your youth might partly determine your socio-economic position in later life (although of course things like coming from a high or low-income family have a huge impact as well). Cultural capital is a broad notion, encompassing not only access to literal culture such as museums, art, books etc., but also having some knowledge of certain social codes (how to speak, how to dress, how to present yourself etc.).

Recently, there has been a lot of interest in the inequality of opportunity between students. For a long time, the view was that if you were smart and persevering enough, you could gain more cultural capital and thus “make something of yourself”, grow and gain a place in a different (read: higher) class than the one you were born in. However, if you don’t have the “appropriate” background knowledge, even working hard might not gain you access to the higher classes. Having access to different forms of cultural capital determines where you’re going in life more often than not. 

Overall, we usually feel like having access to certain “stuff” helps you develop and grow, whether it be intellectual or in your social behavior. If we want to close the inequality gap and assure everyone gets a chance to gain as much cultural capital as possible, shouldn’t everyone have the right to be able to see the new Disney movie or listen to Beyoncé’s new album, even if you or your parents can’t afford to take you to the cinema, subscribe to any and all streaming service out there or go to a museum?

Beneficence, moral duty and civil disobedience

Let’s return to utilitarianism once again. I briefly mentioned that utilitarianists deem something good when it ensures to maximize happiness or welfare, or ensures the most good for the greatest amount of people. To this utilitarian point of view, we might connect the concept of beneficence. From an ethical perspective, this principle refers to a moral obligation. We have a duty to benefit other people and further their interests. 
Let’s return to the example I set in the introduction of this essay. I spoke about getting my students access to a certain tv-series or a novel, but having limited options of doing so. Still, I see the importance of increasing their cultural capital, as I have a wide range of different kids in my classroom. Yes, some have posh parents and will have access to a large bookcase, a subscription to all the different streaming services and regular visits to museums or the theater – but there are also students with parents who barely speak Dutch, or whose parents simply do not have the money to take them places. As a teacher, I feel I have a duty to introduce those students to all the stuff they might not come into contact with at home. I feel in this case, I might be committing an act of both beneficence and civil disobedience when I download a great movie or photocopy a chapter of a book. I am willfully breaching a certain law (downloading the movie instead of taking my students to the cinema, spreading pages of a published work), but I am doing this deliberately, from a feeling of principledness and conscientiousness. 

So let us call upon these principles to justify the infringement of intellectual property rights and using pirated versions of any books, movies, songs or tv-series. I would do this to educate my students and make an attempt to increase their cultural capital and, hopefully, their welfare, even if it’s just the tiniest bit.

Conclusion

I realize that the example of the use of illegal downloads in the classroom might be the most justifiable example there is. We might consent to allow this, but still prohibit private individuals from obtaining illegally downloaded stuff. 

However, a recent study showed that illegal downloading might not be as harmful to (at least the music) industry as the money-making bigshots claim. Getting into contact with the movie of a certain director, the songs of an awesome band or the books of a literary genius might just stimulate us to save up our money, get a ticket to the next show or pre-order their newest novel. Illegal downloading is not just theft, it is a way to discover, explore and get connected with all the awesome stuff that’s out there. And once you’ve done that, there’s always a time or an opportunity to support the makers, thinkers and creators you have learned to love in a way that is possible for you.

This blog was originally written by Isis Hazewindus for the IfThenElse Blogging Collective. It was first posted in March 2022.

Photo credit: Namroud Gorguis via Unsplash

Samuel Butler and Lewis Mumford on technological advancement

“Consciousness, in anything like the present acceptation of the term, having been once a new thing (…) why may not there arise some new phase of mind which shall be as different from all present known phases, as the mind of animals is from that of vegetables? (…) There is no security against the ultimate development of mechanical consciousness, in the fact of machines possessing little consciousness now. (…) reflect upon the extraordinary advance which machines have made during the last few hundred years, and note how slowly the animal and vegetable kingdoms are advancing. (…) what will they not in the end become? Is it not safer to nip the mischief in the bud and to forbid them further progress?”

Above mentioned quote is no recent eureka moment, no modern scientist’s epiphany. The quote is taken from Erewhon, Samuel Butler’s classic utopian novel. 
In 1859 Butler set foot on New-Zealand. The mystical atmosphere of this freshly colonized land and Darwin’s recently published On the Origin of Species would form the breeding ground for Butler’s work, initially as a trilogy of newspaper articles for New-Zealand paper The Press. These articles – one of them aptly named Darwin Among the Machines – formed the basis for the ‘Book of Machines’, and it is the ideas from these chapters that inspired this essay. 

In the next pages, I want to dive into the similarities that can be found in Butler’s Erewhon and an essay by Lewis Mumford on a phenomenon he calls the Megamachine. I will cover both authors’ concerns on technology changing human nature, but I will end on a message of hope Mumford imparts on us in his book on technology in the history of human development.

Butler’s Erewhon

Samuel Butler was not originally a novelist, nor was he a scientist or philosopher. He came from a religious background but became fascinated by Darwin’s theories on biological evolution, which he took to reading at the time he made a journey to New-Zealand to start a sheep-farm. Inspired by Darwin, he started working on some articles and eventually made up a “new doctrine of evolution” in the ensuing book. His newly imagined evolution would not take place in living organisms, but in the iron, steel and coals of the machine – and where better to have this (r)evolution play out, than in a land of imaginary travels, so as to be unhindered by logic and realism. 

Butler made use of the utopia to explore his ideas, and created a remote, idyllic society harboring a great secret. He used his time in New-Zealand as inspiration for his descriptions of Erewhon, the utopian society where the use of all machines is banned – even pocket watches are forbidden.

Butler’s main character is an unknown traveler, whose adventure starts by having him abandon the sheep-farm he works on and risking a dangerous road over the mountains, to find his fortune. When eventually captured by the local people, we discover the land through his eyes.

The traveler quickly notices Erewhon being free of machines, while the inhabitants seem advanced enough. During his first days, one of his attendants discovers his pocket watch, and after this is taken, he is shown a museum of sorts, in which all manner of modern machinery is on display. But it is not until he is more established in Erewhon that he finds out what actually happened to the machines. 

The traveler recounts visiting the land’s university, called the Colleges of Unreason. It is there he is gifted a copy of the book that started the undoing of the machines. In this Book of Machines the writer accounts for the dangers of the machine as a being that might gain consciousness; this piece of text has a definitive Darwinian feel, as there is talk about evolution of consciousness in beings. Consciousness is then explained as a sort of intuitive way of doing things, by anticipating and growing and knowing one’s needs. 

The danger of the machines, according to the revolutionary text, lies in the rapid development of the machines. Human development will stay behind, and the writer doubts even our morality will suffice to “rule” the machines, as he questions ‘the moral sense of any machine’. 

Mumford and the Megamachine

Long after Butler’s musings on evolving machines, in 1966, Lewis Mumford writes the essay Tool Users v.s. Homo Sapiens and the Megamachine:
“Man will create a uniform, all-enveloping structure, designed for operation (…) man will become a passive machine-animal whose proper functions, if this process continues unchanged, will either be fed into a machine or strictly limited and controlled for the benefit of depersonalized collective organizations. The ultimate tendency of this development was correctly anticipated by Samuel Butler, the satirist, more than a century ago: but it is only now that his playful fantasy shows many signs of becoming a far-from-playful reality.”

However, where Butler’s utopian society banned all technology, Mumford emphasizes humankind as a fundamentally technological creature.

Mumford strongly opposes the characterization of humans as a Homo Faber or toolmaker. He finds the idea that humans owe their intelligence to shaping and using tools falling short; he finds the making of tools is nothing special, and argues humans were lagging behind animals in the fabrication of some important utensils such as containers. Instead of Homo Faber, we should characterize humans as Homo Sapiens, or as Mumford also calls it, “mindmakers”.

By characterizing humans as foremostly thinking creatures, Mumford then moves towards defining them as technological creatures, but not in the way we appear to see technology now – as a means to reduce the workload and be as efficient and thus as productive as possible.

Technology, according to Mumford, is an expression of our working brain and a part of our equipment to face life in all its aspects, and not just the cold machine and the ones-and-zeros of the computer. However, if we accept Mumford’s following theory, humans were being ripped from their technological self much longer ago than we might think. 

What we today call technology stems from the emergence of a phenomenon Mumford calls The Megamachine; this is a social, economic or political system in which the cogs in the machine are not made up of iron and steel, but rather of human beings. His most gripping example of the Megamachine is the building of the pyramids of Egypt. The Egyptian Pharaoh amassed hundreds of thousands of slaves to be able to perform the ceaseless work involved in these sublime tombs. According to Mumford, no lifeless machine could have ensured a process as smoothly and efficient as this inexhaustible flow of manpower working day and night, with other slaves in place to step up where one has fallen. Mumford fears the increasing power of the Megamachine in modern times, and part of this fear it will reduce humans to something less than themselves. 

Humans and Robots

Trying to define humans, to specify what makes us unique, we often refer to our rationality and capability of feeling emotions and empathy. No other creature seems to have developed these traits as we have. It is what would place us above the mere animal, but in the last years, the focus of our superiority over the animal kingdom seems to shift to an even more frightful creature: the robot. Built to assist us in any way possible, we keep looking for building even more human-like robots to make them suitable for doing work we now do ourselves. Robots to take care of our elderly should for example be recognizable to make us feel comfortable and not too machine-like. However, this leads to a paradox, because the robot should not be too human, either. 

Our fear of robots is eagerly fed by science-fiction; there are almost no movies or novels in this realm where the robot remains a trusted ally. Most all science-fiction with robots as players end up in a warning: the robot might start to lead its own life, and its urge to efficiency and objectiveness, reinforced by its lack of emotions – which would be impossible to program – makes the robot a dangerous being.

Well before our development of A.I and self-driving cars, Butler describes an ominous future in the Book of Machines, which would come to pass when we do not keep our machines in check. He emphasizes our dependency on machines, and notes that we ‘are bound to them as slaves’. There are workers who give their life tending to the machines, to prevent them from stopping their work and to make sure they continue to live happily.  

This seems eerily current; thinking of developers and programmers designing software and keeping it up to date to ensure continuous accessibility, factories running day and night to supply us with an endless stream of goods we think we need, and ourselves, browsing on our computers and buying every handy new tech-gadget to (as we feel it) improve our lives.

Both Butler and Mumford draw on our fears of the machine, and one could say both see the danger of humans becoming, as Mumford said, a “machine-animal”. But how should we understand this?

Humankind as Machine-Animal

From a Butlerian view, as our use of and dependency on machines grows, two things will happen: humans will become more machine-like, and the machine will become more human-like. To keep our progress going, we will need to build better machines; but this will lead to our lives revolving around consolidating this machine-life. If not, life might turn gruesome: 
“The machines themselves being unable to struggle, have got man to do their struggling for them: as long as he fulfils his function duly, all goes well with him – at least he thinks so; but the moment he fails to do best for the advancement of machinery by encouraging the good and destroying the bad, he is left behind in the race of competition: and this means that he will be made uncomfortable in a variety of ways, and perhaps die.” 
This quote seems to point out a paradox: machines make our lives more comfortable, but we need to keep improving them if we want to stay that way. We work on our increasing our machines’ capacities, while at the same time our day-to-day activities will be reduced to the endless performing of mechanical operations that make us in a way more mechanical than the machines we maintain.

In a way, Butler and Mumford’s view on humankind’s degradation by the “hands” of the machines do not differ too drastically. One of Mumford’s main concerns about our increasing craving for efficiency is that we will, in the end, will be reduced to a “presser of buttons”. But that’s not all Mumford is worried about.

Humans becoming a zombie-like creatures trapped in a system of monotony and repetition is one way of seeing the machine-animal; in an even worse case, we could be separated completely from our natural behaviours. If going on unchecked, the system of the Megamachine would root out what is left of both toolmaker ánd mindmaker. 

Mumford fears that the increasing want for power, production of goods and the yearning for creating infallible and objective intelligence will eventually end humans as we have known them. We will become subservient to the omnipotence and omniscience of the Megamachine.

Curbing the Megamachine 

In the gloom of these predictions, which today seem closer than ever, it could be easy to despair. For what could we do to stop ourselves from our increasing yearning for convenience? Banning all machine-kind but those that are truly analog and unmissable and returning to a simple life, like Butler’s Erewhonians, is not an option – not in the last place because Mumford’s Megamachine is created and sustained by multi-million-dollar companies and governments alike. 

However, we could push ourselves to take action on keeping the might of the Megamachine in check. 

In his book Technics and Human Development – The Myth of the Machine Vol. 1, Mumford explains that we might start an uprising against the Megamachine. In a most successful way, this would be by creating and sustaining the small organizations, such as our universities, societies, or trade unions, that might “rebel” against the vastness of the Megamachine by means of actions, ideas and principles. We need a vast and varied landscape of these, all with a clear and coherent own identity to give some counter-pressure to the danger of losing all individuality and getting sucked up in the bureaucratic system of the Megamachine. Instead of following technical development into a hole of continuous mechanization of our own lives, try to bend our technological development into a way of life that allows us to spend more time on improving ourselves by getting the education we want or do the kind of work that makes us feel satisfied and empowered. So that we, instead of finding ourselves stuck in endless routines and boredom, might truly and meaningfully connect with the toolmaker and the mindmaker once again. 

I would like to end on asking you, dear reader, to keep in mind that there is a world out there full of interesting thoughts and ideas, and that we don’t need a Megamachine to tell us who we are or how to optimize our lives. So attend a lecture in your local library or spend a day reveling in boredom, and remind yourself that sometimes, your own brain is all you need 😉 

This essay was originally written by Isis Hazewindus for the IfThenElse Blogging collective and first posted on November 21^st, 2021.

Literature used for this essay:

Butler, S., Erewhon, Penguin Classics, making use of the chapters The Book of Machines

Mumford, L., Tool users v.s. Homo Sapiens and the Megamachine, from Philosophy of Technology, Wiley Blackwell

Mumford, L., Technics and Human Development – The Myth of the Machine Vol. 1, Harvest/HBJ

Photo credit: Alexandre Debiève on Unsplash