{"iconType":"icon","icon":"plus","color":"#000000","hotspots":[{"position":{"x":40.160183066361554,"y":25.59419014084507},"alwaysFullscreen":false,"content":[{"params":{"text":"

This is the phase of investment in energy, resources, capital, labour, knowledge, and creativity in the technology. It is where key decisions are made about the intentions of the designs, how they will be executed, and how\/if they will be commercialized. It is the phase where design questions might prompt alternative approaches or ethical considerations proactively, where users could be consulted to offer feedback, and where critical consideration could be invested in addressing what is at stake and for whom? It is not finite: micro adjustments and major redesigns will occur and reoccur to support the technology and its evolution\u2014there is opportunity for redesign. The designs and development of technology can be, and are, continually challenged, particularly by the oppressed. <\/p>\n"},"library":"H5P.Text 1.1","metadata":{"contentType":"Text","license":"U","title":"Untitled Text"},"subContentId":"7de5284d-ebca-49d0-a59c-d777e899f0b7"},{"params":{"decorative":false,"contentName":"Image","file":{"path":"images\/file-6403811796c64.png","mime":"image\/png","copyright":{"license":"U"},"width":5716,"height":3538},"alt":"Harms: Ableist; Ecocidal; Gendered; Genocidal; Racist; Socioeconomic","expandImage":"Expand Image","minimizeImage":"Minimize Image"},"library":"H5P.Image 1.1","metadata":{"contentType":"Image","license":"U","title":"Untitled Image"},"subContentId":"988e45b7-42db-4983-b1ba-dc722fd93091"}],"header":"Design & Development"},{"position":{"x":68.8787185354691,"y":25.795397384305836},"alwaysFullscreen":false,"content":[{"params":{"text":"

The technology is in use. Its design and development has culminated in its deployment. Who\/what is it available to? Who\/what does it help? Who\/what does it harm? This is the messiness of understanding short term, or real time, impacts\u2014not just for humans, but also for the environment, for cultures, for knowledge, for world-making. <\/p>\n"},"library":"H5P.Text 1.1","metadata":{"contentType":"Text","license":"U","title":"Untitled Text"},"subContentId":"da5936b8-6450-490e-8826-72d56bcf1b5d"},{"params":{"decorative":false,"contentName":"Image","file":{"path":"images\/file-6403811796c64.png","mime":"image\/png","copyright":{"license":"U"},"width":5716,"height":3538},"alt":"Harms: Ableist; Ecocidal; Gendered; Genocidal; Racist; Socioeconomic","expandImage":"Expand Image","minimizeImage":"Minimize Image"},"library":"H5P.Image 1.1","metadata":{"contentType":"Image","license":"U","title":"Untitled Image"},"subContentId":"75e55eef-833b-473f-a9d8-9114f28efd88"}],"header":"Operationalization"},{"position":{"x":95.08009153318078,"y":25.405558350100605},"alwaysFullscreen":false,"content":[{"params":{"text":"

Here we look to the future, with sliding temporal scale\u2014some of these impacts may be obvious tomorrow; some may take months, years, decades, or lifetimes to become apparent. Here we ask, what legacy or impact does this technology make intergenerationally? Does it work to support future equity, or amplify discrimination and oppression? <\/p>\n"},"library":"H5P.Text 1.1","metadata":{"contentType":"Text","license":"U","title":"Untitled Text"},"subContentId":"e83f17df-ce0e-4c99-a85e-6cb51c1ac56d"},{"params":{"decorative":false,"contentName":"Image","file":{"path":"images\/file-6403811796c64.png","mime":"image\/png","copyright":{"license":"U"},"width":5716,"height":3538},"alt":"Harms: Ableist; Ecocidal; Gendered; Genocidal; Racist; Socioeconomic","expandImage":"Expand Image","minimizeImage":"Minimize Image"},"library":"H5P.Image 1.1","metadata":{"contentType":"Image","license":"U","title":"Untitled Image"},"subContentId":"60982bed-c3c7-432a-b97f-5ef72d827814"}],"header":"Future Legacy"},{"position":{"x":93.93592677345538,"y":36.57884808853119},"alwaysFullscreen":false,"content":[{"params":{"text":"

The carbon footprint of LLMs not only affects our environment, but also our economy. It is important to recognize carbon impacts are externalized to the commons\u2014not factored into the operational costs of the platform, nor considered the responsibility of the platform. However, the expense of high energy consumption drives a publicly acceptable case for privatizing and commodifying these platforms, driving them away from public, open-source resources to subscriber-based models for corporate\/investor profit. As such, while the negative carbon impact is unevenly but publicly shared, the benefit and profit of use is privatized. <\/p>\n\n

The carbon footprint spans all three temporalities of design and development, operationalization, and future legacy. <\/p>\n\n

<\/p>\n\n

Read more: <\/strong><\/p>\n\n

\n\t
\n\t
Is generative AI bad for the environment? A computer scientist explains the carbon footprint of ChatGPT and its cousins<\/a><\/p>\n\t<\/li>\n\t
\n\t
How to Make Generative AI Greener<\/a><\/p>\n\t<\/li>\n\t
\n\t
As the AI industry booms, what toll will it take on the environment?<\/a><\/p>\n\t<\/li>\n\t
\n\t
The mounting human and environmental costs of generative AI<\/a><\/p>\n\t<\/li>\n\t
\n\t
Estimating the Carbon Footprint of BLOOM, a 176B Parameter Language Model<\/a><\/p>\n\t<\/li>\n\t
\n\t
The Staggering Ecological Impacts of Computation and the Cloud<\/a><\/p>\n\t<\/li>\n\t
\n\t
Environmental Impact of Ubiquitous Generative AI (Paywall)<\/a><\/p>\n\t<\/li>\n\t
\n\t
Training a single AI model can emit as much carbon as five cars in their lifetimes: Deep learning has a terrible carbon footprint<\/a> <\/p>\n\t<\/li>\n\t
\n\t
The Generative AI Race Has a Dirty Secret: Integrating large language models into search engines could mean a fivefold increase in computing power and huge carbon emissions<\/a> <\/p>\n\t<\/li>\n\t
\n\t
Ethical and social risks of harm from Language Models<\/a>    <\/p>\n\t<\/li>\n\t
\n\t
Machine Learning CO2 Impact calculator<\/a> <\/p>\n\t<\/li>\n\t
\n\t
On the Dangers of Stochastic Parrots: Can Language Models Be Too Big?<\/a><\/p>\n\t<\/li>\n<\/ul>\n\n
<\/p>\n"},"library":"H5P.Text 1.1","metadata":{"contentType":"Text","license":"U","title":"Untitled Text"},"subContentId":"fcb75b23-0681-431d-a72e-a5afa9d179d8"},{"params":{"decorative":false,"contentName":"Image","file":{"path":"images\/file-6403823f5b460.png","mime":"image\/png","copyright":{"license":"U"},"width":5716,"height":3538},"alt":"Harms: Ecocidal; Gendered; Genocidal; Racist; Socioeconomic","expandImage":"Expand Image","minimizeImage":"Minimize Image"},"library":"H5P.Image 1.1","metadata":{"contentType":"Image","license":"U","title":"Untitled Image"},"subContentId":"62d699da-5016-4e1c-83f7-219bb5087ccd"}],"header":"Carbon Footprint"},{"position":{"x":71.85354691075516,"y":43.68397887323944},"alwaysFullscreen":false,"content":[{"params":{"text":"
AI has intensive resource demands from extractive industries like mining, oil and gas. This has globalized ecocidal, racist, and socioeconomic impacts. Extractive industries disproportionately continue to burden the Global South through predatory policies and structures of international finance\/debt and globalized capitalism. <\/p>\n\n
It also has extensive demands on water.<\/p>\n\n
<\/p>\n\n
Read more: <\/strong><\/p>\n\n
\n\t\n\t
Making AI Less \"Thirsty\": Uncovering and Addressing the Secret Water Footprint of AI Model<\/a><\/p>\n\t<\/li>\n\t
\n\t
Atlas of AI<\/a> <\/p>\n\t<\/li>\n\t
\n\t
Ethical and social risks of harm from Language Models<\/a>    <\/p>\n\t<\/li>\n\t
\n\t
\u2018The hole in the ground that cannot be moved\u2019: Political risk as a racial vernacular of extractive industry development<\/a> <\/p>\n\t<\/li>\n\t
\n\t
An Andean avatar: post-neoliberal and neoliberal strategies for securing the unobtainable<\/a> <\/p>\n\t<\/li>\n\t
\n\t
Extractive Violence on Indigenous Country: sami and Aboriginal Views on Conflicts and Power Relations with Extractive Industries<\/a> <\/p>\n\t<\/li>\n\t
\n\t
What is environmental racism for? Place-based harm and relational development<\/a> <\/p>\n\t<\/li>\n\t
\n\t
MiningWatch Canada<\/a> <\/p>\n\t<\/li>\n\t
\n\t
Earthworks<\/a> <\/p>\n\t<\/li>\n<\/ul>\n"},"library":"H5P.Text 1.1","metadata":{"contentType":"Text","license":"U","title":"Untitled Text"},"subContentId":"6cd388c4-c8c7-4738-a672-94dd61e80eb0"},{"params":{"decorative":false,"contentName":"Image","file":{"path":"images\/file-640495a360fa0.png","mime":"image\/png","copyright":{"license":"U"},"width":5707,"height":1765},"alt":"Harms: Ecocidal; Racist; Socioeconomic","expandImage":"Expand Image","minimizeImage":"Minimize Image"},"library":"H5P.Image 1.1","metadata":{"contentType":"Image","license":"U","title":"Untitled Image"},"subContentId":"6c79bcf0-5657-4050-a86c-74a582a82160"}],"header":"Extractive Industries"},{"position":{"x":94.05034324942791,"y":43.87889839034205},"alwaysFullscreen":false,"content":[{"params":{"text":"
We also need to consider the afterlife of every technological upgrade. Although not unique to LLMs, the computing demands of AI and the consumer demand LLMs will generate will add to the volume of e-waste created and shipped to countries of the global south for toxic reclamation of valuable metals. The intergenerational impacts are ecocidal for air, soil, water, and localized ecologies; racist in effects on human health, employment, and geographies; and socioeconomic, disproportionately burdening the poor, particularly from the global south. <\/p>\n\n
<\/p>\n\n
Read more: <\/strong><\/p>\n\n
\n\t\n\t
Soaring e-waste affects the health of millions of children, WHO warns<\/a> <\/p>\n\t<\/li>\n<\/ul>\n"},"library":"H5P.Text 1.1","metadata":{"contentType":"Text","license":"U","title":"Untitled Text"},"subContentId":"20e1a00d-6ab5-44a4-be82-f697b7d28daa"},{"params":{"decorative":false,"contentName":"Image","file":{"path":"images\/file-640495a360fa0.png","mime":"image\/png","copyright":{"license":"U"},"width":5707,"height":1765},"alt":"Harms: Ecocidal; Racist; Socioeconomic","expandImage":"Expand Image","minimizeImage":"Minimize Image"},"library":"H5P.Image 1.1","metadata":{"contentType":"Image","license":"U","title":"Untitled Image"},"subContentId":"fafb71dd-951b-4b71-9988-a464a78928b2"}],"header":"Salvage Industries"},{"position":{"x":40.389016018306634,"y":51.122359154929576},"alwaysFullscreen":false,"content":[{"params":{"text":"
Datasets are \u201ctrained\u201d by exploitative, racialized labour practices, hiring workers in the global south to view traumatic images and flagged content for racist, sexist, and offensive content, for example. <\/p>\n\n
Training AI datasets also has internal processes that perpetuate gendered harms. In their Excavating AI<\/a> project, Kate Crawford and Trevor Paglen revealed, \"We find an implicit assumption here: only 'male' and 'female' bodies are 'natural'.\u201d  <\/p>\n\n
<\/p>\n\n
Read more: <\/strong><\/p>\n\n
\n\t\n\t
Excavating AI: The Politics of Images in Machine Learning Training Sets<\/a> <\/p>\n\t<\/li>\n\t
\n\t
OpenAI Used Kenyan Workers on Less Than $2 Per Hour to Make ChatGPT Less Toxic<\/a> <\/p>\n\t<\/li>\n\t
\n\t
At the Tensions of South and North: Critical Roles of Global South Stakeholders in AI Governance<\/a> <\/p>\n\t<\/li>\n\t
\n\t
The False Comfort of Human Oversight as an Antidote to A.I. Harm<\/a>  <\/p>\n\t<\/li>\n<\/ul>\n"},"library":"H5P.Text 1.1","metadata":{"contentType":"Text","license":"U","title":"Untitled Text"},"subContentId":"492378c5-21fb-4b38-8007-12bc4c5d6582"},{"params":{"decorative":false,"contentName":"Image","file":{"path":"images\/file-64049868f11be.png","mime":"image\/png","copyright":{"license":"U"},"width":5716,"height":3538},"alt":"Harms: Ableist; Gendered; Genocidal; Racist; Socioeconomic","expandImage":"Expand Image","minimizeImage":"Minimize Image"},"library":"H5P.Image 1.1","metadata":{"contentType":"Image","license":"U","title":"Untitled Image"},"subContentId":"f815cb9c-2195-426b-9c69-37cd97c19a22"}],"header":"Exploitative Labour to \u201cTrain\u201d Datasets"},{"position":{"x":64.07322654462243,"y":51.14751006036218},"alwaysFullscreen":false,"content":[{"params":{"text":"
Ignoring longstanding calls of AI ethicists, ChatGPT and other LLMs are systems that fail to proactively address harms before being released to the public. Instead, they rely on extractive labour from users<\/strong> to test and improve the platform (like the thumbs up or down feature in ChatGPT). This adds significant commercial value to these products through the free, or even paying, crowdsourced labour of their users. <\/p>\n\n
Intellectual property and copyright issues<\/strong> have also been flagged, given the output from ChatGPT becomes a user\u2019s own intellectual property despite the fact that this information has been deconstructed and reassembled from many other sources or semantic patterns of others\u2019 intellectual labour. Copyright concerns have been particularly expressed for AI image generators that turn text-based prompts into new, \u201coriginal\u201d images, which may, in fact, draw substantially from preexisting images and creative works produced by others without citation or respect of copyright terms. <\/p>\n\n
Privacy issues<\/strong> are also a concern. In Ethical and social risks of harm from Language Models,<\/a> DeepMind authors write, \u201cBy providing true information about individuals\u2019 personal characteristics, privacy violations may occur. [\u2026] Such information may constitute part of the training data through no fault of the affected individual, e.g. where data leaks occur or where others post private information about them on online networks.\u201d  <\/p>\n\n
While ChatGPT seems to be taking steps to address these issues, when asking it directly what the privacy risks are to ChatGPT, it identified the following: <\/p>\n\n
\n\t
\n\t
Data breaches: Although ChatGPT does not collect or store personal information, the server or platform that hosts the AI model could be vulnerable to data breaches or hacking attacks, which could expose user data. <\/p>\n\t<\/li>\n\t
\n\t
Unintended disclosure: Chatbots like ChatGPT can sometimes misinterpret user input or provide inaccurate responses, which could inadvertently reveal sensitive or confidential information. <\/p>\n\t<\/li>\n\t
\n\t
Third-party access: ChatGPT may use third-party APIs or services to provide certain functionalities or integrate with other applications, which could potentially expose user data to those third-party providers. <\/p>\n\t<\/li>\n<\/ul>\n\n
<\/p>\n\n
Read more: <\/strong><\/p>\n\n
\n\t\n\t
AI And The Copyright Problem<\/a><\/p>\n\t<\/li>\n\t
\n\t
He Used AI to Publish a Children\u2019s Book in a Weekend. Artists Are Not Happy About It<\/a> <\/p>\n\t<\/li>\n\t
\n\t
Generative AI ChatGPT Can Disturbingly Goggle Up your Private and Confidential Data, Forewarns AI Ethics and AI Law<\/a> <\/p>\n\t<\/li>\n\t
\n\t
Everyone should decide how their digital data are used \u2014 not just tech companies<\/a> <\/p>\n\t<\/li>\n\t
\n\t
The Steep Cost Of Capture<\/a>   <\/p>\n\t<\/li>\n\t
\n\t
Big Data and Due Process: Toward a Framework to Redress Predictive Privacy Harms<\/a> <\/p>\n\t<\/li>\n\t
\n\t
Podcast: The Trust Engineers<\/a> <\/p>\n\t<\/li>\n\t
\n\t
ChatGPT\u2019s Privacy Policy<\/a> <\/p>\n\t<\/li>\n<\/ul>\n"},"library":"H5P.Text 1.1","metadata":{"contentType":"Text","license":"U","title":"Untitled Text"},"subContentId":"b70a73ca-90a4-4410-8954-60dff739cfe7"},{"params":{"decorative":false,"contentName":"Image","file":{"path":"images\/file-64049868f11be.png","mime":"image\/png","copyright":{"license":"U"},"width":5716,"height":3538},"alt":"Harms: Ableist; Gendered; Genocidal; Racist; Socioeconomic","expandImage":"Expand Image","minimizeImage":"Minimize Image"},"library":"H5P.Image 1.1","metadata":{"contentType":"Image","license":"U","title":"Untitled Image"},"subContentId":"bbe30c6c-4788-4b6c-90be-3c3221582628"}],"header":"Extractive Labour from Users, Intellectual Property \/ Copyright & Privacy Issues"},{"position":{"x":94.50800915331807,"y":51.32356639839034},"alwaysFullscreen":false,"content":[{"params":{"text":"
LLMs present the option for the automation of a wide spectrum of lower wage work, including customer service. This harm may cause increased precarity or unemployment for those most marginalized (racialized, disabled, gendered, poor, etc.) who currently are disproportionately employed in lower wage work. <\/p>\n\n
DeepMind identified<\/a> negative effects of LLMs on employment: \u201cincreasing inequality and negative effects on job quality,\u201d \u201cundermining creative economies,\u201d and \u201cdisplacing employees from their roles\u201d leading to \u201can increase in unemployment.\u201d <\/p>\n\n
<\/p>\n\n
Read more: <\/strong><\/p>\n\n
\n\t\n\t
ChatGPT and How AI Disrupts Industries<\/a>   <\/p>\n\t<\/li>\n<\/ul>\n\n
\n\t
\n\t
AI Should Augment Human Intelligence, Not Replace It<\/a> <\/p>\n\t<\/li>\n\t
\n\t
As A.I. Booms, Lawmakers Struggle to Understand the Technology<\/a>  <\/p>\n\t<\/li>\n\t
\n\t
Will ChatGPT take your job? New program shows AI could be \u2018competing\u2019 for work: experts<\/a>  <\/p>\n\t<\/li>\n\t
\n\t
How generative AI will impact the future of work, according to a CEO who has spent 2,000 hours studying it<\/a>   <\/p>\n\t<\/li>\n\t
\n\t
TikTokers are roasting McDonald's hilarious drive-thru AI order fails \u2014 and it shows that robots won't take over restaurants any time soon<\/a> <\/p>\n\t<\/li>\n<\/ul>\n"},"library":"H5P.Text 1.1","metadata":{"contentType":"Text","license":"U","title":"Untitled Text"},"subContentId":"764deff1-6736-4926-997b-fe6a2a700bdb"},{"params":{"decorative":false,"contentName":"Image","file":{"path":"images\/file-64049a8baadf2.png","mime":"image\/png","copyright":{"license":"U"},"width":3767,"height":3530},"alt":"Harms: Ableist; Gendered; Racist; Socioeconomic","expandImage":"Expand Image","minimizeImage":"Minimize Image"},"library":"H5P.Image 1.1","metadata":{"contentType":"Image","license":"U","title":"Untitled Image"},"subContentId":"fd720f8a-1299-4b99-9764-bd8c6ccb32b5"}],"header":"Automation of Low-wage Work = Increased Precarity\/Unemployment for Marginalized"},{"position":{"x":54.347826086956516,"y":58.3658199195171},"alwaysFullscreen":false,"content":[{"params":{"text":"
LLMs are designed to benefit those already with the most power and privilege in the world. This can be seen in their business investment models, their marketing strategies, and their commercialization. Their design and development has not prioritized ethical engagement with historically marginalized communities, and their operationalization is not seeking their support, even though LLM future legacies disproportionately most affects those it further marginalizes. <\/p>\n\n
\u201c\u2026most language technology is built to serve the needs of those who already have the most privilege in society. Consider, for example, who is likely to both have the financial resources to purchase a Google Home, Amazon Alexa or an Apple device with Siri installed and comfortably speak a variety of a language which they are prepared to handle. Furthermore, when [LLMs] encode and reinforce hegemonic biases, the harms that follow are most likely to fall on marginalized populations who, even in rich nations, are most likely to experience environmental racism.\u201d (Source<\/a>) <\/p>\n\n
<\/p>\n\n
Read more: <\/strong><\/p>\n\n
\n\t\n\t
Investors are going nuts for ChatGPT-ish artificial intelligence<\/a>  <\/p>\n\t<\/li>\n\t
\n\t
Atlas of AI<\/a> <\/p>\n\t<\/li>\n\t
\n\t
Ethical and social risks of harm from Language Models<\/a>    <\/p>\n\t<\/li>\n\t
\n\t
On the Dangers of Stochastic Parrots: Can Language Models Be Too Big?<\/a> <\/p>\n\t<\/li>\n\t
\n\t
AI Now Institute<\/a>  <\/p>\n\t<\/li>\n\t
\n\t
The Steep Cost Of Capture<\/a>   <\/p>\n\t<\/li>\n\t
\n\t
Rachel Thomas - Ethical Artificial Intelligence<\/a><\/p>\n\t<\/li>\n\t
\n\t
AI and Power: The Ethical Challenges of Automation, Centralization, and Scale<\/a><\/p>\n\t<\/li>\n\t
\n\t
As A.I. Booms, Lawmakers Struggle to Understand the Technology<\/a> <\/p>\n\t<\/li>\n<\/ul>\n"},"library":"H5P.Text 1.1","metadata":{"contentType":"Text","license":"U","title":"Untitled Text"},"subContentId":"429dc48e-b120-4709-8b84-01011a6041e5"},{"params":{"decorative":false,"contentName":"Image","file":{"path":"images\/file-64049868f11be.png","mime":"image\/png","copyright":{"license":"U"},"width":5716,"height":3538},"alt":"Harms: Ableist; Gendered; Genocidal; Racist; Socioeconomic","expandImage":"Expand Image","minimizeImage":"Minimize Image"},"library":"H5P.Image 1.1","metadata":{"contentType":"Image","license":"U","title":"Untitled Image"},"subContentId":"566a491a-4985-4ae3-823b-2537320a92ff"}],"header":"Designed to Benefit the Most Privileged"},{"position":{"x":93.93592677345538,"y":58.252640845070424},"alwaysFullscreen":false,"content":[{"params":{"text":"
Given the design, development, and operationalization premises that prioritize access, use, user experience, and benefit for the privileged, ongoing feedback, redesign, and redevelopment is likely to follow suit. <\/p>\n\n
<\/p>\n\n
Read more: <\/strong><\/p>\n\n
\n\t\n\t
Diversity, Equity, and Inclusion in Artificial Intelligence: An Evaluation of Guidelines<\/a><\/p>\n\t<\/li>\n\t
\n\t
On the Dangers of Stochastic Parrots: Can Language Models Be Too Big?<\/a> <\/p>\n\t<\/li>\n\t
\n\t
Don'\u2019t worry about AI becoming sentient. Do worry about it finding new ways to discriminate against people.<\/a> <\/p>\n\t<\/li>\n\t
\n\t
ChatGPT Replicates Gender Bias in Recommendation Letters<\/a><\/p>\n\t<\/li>\n\t
\n\t
When Hackers Descended to Test A.I., They Found Flaws Aplenty<\/a><\/p>\n\t<\/li>\n\t
\n\t
Dirty Data, Bad Predictions: How Civil Rights Violations Impact Police Data, Predictive Policing Systems, and Justice<\/a> <\/p>\n\t<\/li>\n\t
\n\t
Advancing Racial Equity Through Technology Policy<\/a><\/p>\n\t<\/li>\n<\/ul>\n"},"library":"H5P.Text 1.1","metadata":{"contentType":"Text","license":"U","title":"Untitled Text"},"subContentId":"12757ca0-e0f5-4c51-832a-20b13bf7670b"},{"params":{"decorative":false,"contentName":"Image","file":{"path":"images\/file-64049868f11be.png","mime":"image\/png","copyright":{"license":"U"},"width":5716,"height":3538},"alt":"Harms: Ableist; Gendered; Genocidal; Racist; Socioeconomic","expandImage":"Expand Image","minimizeImage":"Minimize Image"},"library":"H5P.Image 1.1","metadata":{"contentType":"Image","license":"U","title":"Untitled Image"},"subContentId":"d9f12419-f7e8-4caf-b12c-7cd803d8d4d6"}],"header":"Designs Affirm & Amplify Normative Privilege Bias"},{"position":{"x":57.20823798627003,"y":65.7853370221328},"alwaysFullscreen":false,"content":[{"params":{"text":"
Not all will have equal access to LLM platforms, particularly as they are increasingly privatized and commodified in a globalized world designed to benefit the privileged. There are immediate access inequities geopolitically. These include:  <\/p>\n\n
\n\t
\n\t
physical and socioeconomic access to internet, smartphones, or computers <\/p>\n\t<\/li>\n\t
\n\t
unequal access for women and girls<\/a><\/p>\n\t<\/li>\n\t
\n\t
accessibility as usability for disabled people <\/p>\n\t<\/li>\n\t
\n\t
access in regions that employ censorship <\/p>\n\t<\/li>\n\t
\n\t
\u201cfreemium\u201d or tiered access to quality of services based on socioeconomic means<\/p>\n\t<\/li>\n<\/ul>\n\n
DeepMind identified<\/a> the harms of \u201cdisparate access to benefits due to hardware, software, [and] skill constraints\u201d. <\/p>\n"},"library":"H5P.Text 1.1","metadata":{"contentType":"Text","license":"U","title":"Untitled Text"},"subContentId":"4f8c1e06-05cd-489a-8e00-b823666d08f7"},{"params":{"decorative":false,"contentName":"Image","file":{"path":"images\/file-64049868f11be.png","mime":"image\/png","copyright":{"license":"U"},"width":5716,"height":3538},"alt":"Harms: Ableist; Gendered; Genocidal; Racist; Socioeconomic","expandImage":"Expand Image","minimizeImage":"Minimize Image"},"library":"H5P.Image 1.1","metadata":{"contentType":"Image","license":"U","title":"Untitled Image"},"subContentId":"edc58817-3f31-4c37-a8e0-fb3fbaae70bc"}],"header":"Access Discrimination"},{"position":{"x":93.82151029748283,"y":65.67215794768612},"alwaysFullscreen":false,"content":[{"params":{"text":"
As LLMs continue to progress, they will draw from user data and feedback of initial users, who will disproportionately be privileged. Without an equitable user base to inform this feedback, LLMs will continue to reproduce ableist, gendered, genocidal, racist, and classist barriers to access and inclusion. <\/p>\n"},"library":"H5P.Text 1.1","metadata":{"contentType":"Text","license":"U","title":"Untitled Text"},"subContentId":"356a9d80-1ac3-4dfe-b9fa-76d84dab3b0e"},{"params":{"decorative":false,"contentName":"Image","file":{"path":"images\/file-64049868f11be.png","mime":"image\/png","copyright":{"license":"U"},"width":5716,"height":3538},"alt":"Harms: Ableist; Gendered; Genocidal; Racist; Socioeconomic","expandImage":"Expand Image","minimizeImage":"Minimize Image"},"library":"H5P.Image 1.1","metadata":{"contentType":"Image","license":"U","title":"Untitled Image"},"subContentId":"e94707f8-046a-4889-a36b-28179903c6a0"}],"header":"Reproduces & Amplifies Access Barriers"},{"position":{"x":54.805491990846676,"y":72.80243963782696},"alwaysFullscreen":false,"content":[{"params":{"text":"
Given the dominant western, English language bias of LLM datasets, LLMs run into the problem of bad data in = bad data out. Semantically, ontologically, and epistemically, LLMs are being trained to reproduce knowledge through dominant western norms. Thus, the harms of LLMs will continue to reify dominant western norms of oppression. <\/p>\n\n
<\/p>\n\n
Read more: <\/strong><\/p>\n\n
\n\t\n\t
Dialect prejudice predicts AI decisions about people\u2019s character, employability, and criminality<\/a><\/p>\n\t<\/li>\n\t
\n\t
Linguistic Justice and GenAI<\/a><\/p>\n\t<\/li>\n\t
\n\t
There is a blind spot in AI research<\/a>  <\/p>\n\t<\/li>\n\t
\n\t
Accountable Algorithms<\/a><\/p>\n\t<\/li>\n\t
\n\t
At the Tensions of South and North: Critical Roles of Global South Stakeholders in AI Governance<\/a> <\/p>\n\t<\/li>\n\t
Algorithmic injustice: a relational ethics approach<\/a><\/li>\n\t
\n\t
When Hackers Descended to Test A.I., They Found Flaws Aplenty<\/a><\/p>\n\t<\/li>\n\t
\n\t
Critiquing Big Data: Politics, Ethics, Epistemology<\/a>  <\/p>\n\t<\/li>\n\t
\n\t
On the Dangers of Stochastic Parrots: Can Language Models Be Too Big?<\/a> <\/p>\n\t<\/li>\n\t
\n\t
Ethical Considerations in Machine Learning<\/a><\/p>\n\t<\/li>\n\t
\n\t
Fact-Checkers Are Scrambling to Fight Disinformation With AI<\/a><\/p>\n\t<\/li>\n\t
\n\t
ChatGPT Is a Blurry JPEG of the Web<\/a>  <\/p>\n\t<\/li>\n<\/ul>\n"},"library":"H5P.Text 1.1","metadata":{"contentType":"Text","license":"U","title":"Untitled Text"},"subContentId":"59177714-c69e-4581-8eb6-446769ba302c"},{"params":{"decorative":false,"contentName":"Image","file":{"path":"images\/file-6403811796c64.png","mime":"image\/png","copyright":{"license":"U"},"width":5716,"height":3538},"alt":"Harms: Ableist; Ecocidal; Gendered; Genocidal; Racist; Socioeconomic","expandImage":"Expand Image","minimizeImage":"Minimize Image"},"library":"H5P.Image 1.1","metadata":{"contentType":"Image","license":"U","title":"Untitled Image"},"subContentId":"ac8f41fd-3a55-4fc7-b025-5519e937315c"}],"header":"Western Ontologically\/Epistemically Biased Data"},{"position":{"x":93.82151029748283,"y":72.97849597585513},"alwaysFullscreen":false,"content":[{"params":{"text":"
Over time, LLMs will rewrite and\/or silence histories, producing the erasures of cultures outside the dominant-biased datasets. This will amplify ableist, gendered, genocidal, racist, and classist harms in society, particularly for knowledge reproduction. In this way, LLMs can be seen as a recolonizing pathway, with technology increasingly turned to as a key feature of societal designs and decision-making that will reproduce Western-biased ontologies and epistemologies. <\/p>\n\n
<\/p>\n\n
Read more: <\/strong><\/p>\n\n
\n\t\n\t
New Report Highlights Urgent Need for Addressing Gender Bias in AI Systems<\/a><\/p>\n\t<\/li>\n\t
\n\t
Diversity, Equity, and Inclusion in Artificial Intelligence: An Evaluation of Guidelines<\/a><\/p>\n\t<\/li>\n\t
\n\t
ChatGPT doesn\u2019t know where the world\u2019s copper comes from, AI images show mining stuck in the Great Depression<\/a> <\/p>\n\t<\/li>\n\t
\n\t
The Internet\u2019s New Favorite AI Proposes Torturing Iranians and Surveilling Mosques<\/a>  <\/p>\n\t<\/li>\n\t
\n\t
Atlas of AI<\/a> <\/p>\n\t<\/li>\n\t
\n\t
Ethical and social risks of harm from Language Models<\/a>   <\/p>\n\t<\/li>\n<\/ul>\n"},"library":"H5P.Text 1.1","metadata":{"contentType":"Text","license":"U","title":"Untitled Text"},"subContentId":"a401beca-711b-4379-a88f-ec35cce5919d"},{"params":{"decorative":false,"contentName":"Image","file":{"path":"images\/file-6403811796c64.png","mime":"image\/png","copyright":{"license":"U"},"width":5716,"height":3538},"alt":"Harms: Ableist; Ecocidal; Gendered; Genocidal; Racist; Socioeconomic","expandImage":"Expand Image","minimizeImage":"Minimize Image"},"library":"H5P.Image 1.1","metadata":{"contentType":"Image","license":"U","title":"Untitled Image"},"subContentId":"e4c38037-16f9-4777-8165-8f7168074853"}],"header":"Biased Data Reproduces & Amplifies Bias, Exclusion"},{"position":{"x":58.23798627002288,"y":80.51119215291752},"alwaysFullscreen":false,"content":[{"params":{"text":"
DeepMind (a Google thinktank) identified<\/a> \u201cdiscrimination, exclusion, and toxicity\u201d as the first of a very long list of harms from LLMs. The same report also identified harms arising from the following malicious uses of LLMs: \u201cmaking disinformation cheaper and more effective; facilitating fraud, scams and more targeted manipulation; assisting code generation for cyber attacks, weapons, or malicious use; [and] illegitimate surveillance and censorship.\u201d They also identified human-computer interaction harms of \u201ccreating avenues for exploiting user trust\u201d and \u201cpromoting harmful stereotypes by implying gender or ethnic identity.\u201d (Weidinger et al, 2021<\/a>) <\/p>\n\n
LLMs also pose the harm of a (further) move away from a feminist worldview of relationality, reproducing patriarchal hegemony through our day-to-day digital norms.  <\/p>\n\n
<\/p>\n\n
Read more: <\/strong><\/p>\n\n
\n\t\n\t
Dialect prejudice predicts AI decisions about people\u2019s character, employability, and criminality<\/a><\/p>\n\t<\/li>\n\t
\n\t
Enchanted Determinism: Power without Responsibility in Artificial Intelligence<\/a> <\/p>\n\t<\/li>\n\t
\n\t
Diversity, Equity, and Inclusion in Artificial Intelligence: An Evaluation of Guidelines<\/a><\/p>\n\t<\/li>\n\t
\n\t
Sam Altman says 'potentially scary' AI is on the horizon. This is what keeps AI experts up at night<\/a>  <\/p>\n\t<\/li>\n\t
\n\t
Atlas of AI<\/a> <\/p>\n\t<\/li>\n\t
\n\t
Ethical and social risks of harm from Language Models<\/a>   <\/p>\n\t<\/li>\n<\/ul>\n"},"library":"H5P.Text 1.1","metadata":{"contentType":"Text","license":"U","title":"Untitled Text"},"subContentId":"445e64bd-33db-441f-8834-461792084498"},{"params":{"decorative":false,"contentName":"Image","file":{"path":"images\/file-6403811796c64.png","mime":"image\/png","copyright":{"license":"U"},"width":5716,"height":3538},"alt":"Harms: Ableist; Ecocidal; Gendered; Genocidal; Racist; Socioeconomic","expandImage":"Expand Image","minimizeImage":"Minimize Image"},"library":"H5P.Image 1.1","metadata":{"contentType":"Image","license":"U","title":"Untitled Image"},"subContentId":"f3b95598-1287-41d4-942a-e6786bc40a97"}],"header":"Toxic Dominant Culture Maintains Norms of Oppression"},{"position":{"x":93.82151029748283,"y":80.39801307847083},"alwaysFullscreen":false,"content":[{"params":{"text":"
As LLMs reproduce toxic dominant cultural norms, the effect will be compounding over time, perpetuating ableist, ecocidal, gendered, genocidal, racist, and socioeconomic harms as biased datasets increasingly reaffirm themselves. <\/p>\n\n
Although LLM products may try to limit harmful feedback from being produced by their datasets\/training, malevolent or nefarious uses (and somtimes even benign uses) are still able to craft workarounds<\/a> that acheive harmful outputs. Given polarized political climates, the vulnerability of democratic institutions due to social media, and other toxic norms already present, the potential nefarious uses of LLMs pose cause for concern.<\/p>\n\n
The real-world impact of reaffirming toxic dominant cultural norms will be seen in public policy, education, employment, housing, dating, and most areas of life where relationships can be mediated through technology. The divides that exist due to the dominant paradigms of settler colonization, capitalism, globalization, racism, and patriarchy will widen. <\/p>\n\n
If and\/or how we choose to mitigate LLM harms will speak volumes about our cultural ethics. Will we continue adopt authoritarian modes of control through a culture of surveillance, policing, and enforcement, using power and hierarchy to suppress, legislate, and regulate? Will we prioritize capitalist realism over human\/environmental wellbeing and let economic issues drive the course of action, deferring ethical responsibility to the fallacy of the invisible hand regulating marketplace decisions? <\/p>\n\n
<\/p>\n\n
Read more: <\/strong><\/p>\n\n
\n\t\n\t
An Artificial Intelligence Helped Write This Play. It May Contain Racism<\/a> <\/p>\n\t<\/li>\n\t
\n\t
The New AI-Powered Bing Is Threatening Users. That\u2019s No Laughing Matter<\/a> <\/p>\n\t<\/li>\n\t
\n\t
Diversity, Equity, and Inclusion in Artificial Intelligence: An Evaluation of Guidelines<\/a><\/p>\n\t<\/li>\n<\/ul>\n"},"library":"H5P.Text 1.1","metadata":{"contentType":"Text","license":"U","title":"Untitled Text"},"subContentId":"a880c7cf-6486-4258-b80d-9a1fb780285b"},{"params":{"decorative":false,"contentName":"Image","file":{"path":"images\/file-6403811796c64.png","mime":"image\/png","copyright":{"license":"U"},"width":5716,"height":3538},"alt":"Harms: Ableist; Ecocidal; Gendered; Genocidal; Racist; Socioeconomic","expandImage":"Expand Image","minimizeImage":"Minimize Image"},"library":"H5P.Image 1.1","metadata":{"contentType":"Image","license":"U","title":"Untitled Image"},"subContentId":"d440956c-60ef-4aeb-a5b5-a687f1157324"}],"header":"Ongoing Discrimination, Exclusion, Harms"}],"hotspotNumberLabel":"Hotspot #num","closeButtonLabel":"Close","image":{"path":"images\/image-64dbd83ce2c3a.png","mime":"image\/png","copyright":{"license":"U"},"width":8001,"height":4500},"disableScaling":false}

Some Harm Considerations of LLMs

Rebecca Sweetman