6 Matching Annotations
  1. Last 7 days
  2. markcmarino.com markcmarino.com
    parrots.pdf
    6
    1. typicallyashj 26 May 2026
      Jones [ 64] documents (using digital ethnography techniques [ 63])multiple cases where people on the receiving end of death threatson Twitter have had their accounts suspended while the accountsissuing the death threats persist. She further reports that harass-ment on Twitter is experienced by “a wide range of overlappinggroups including domestic abuse victims, sex workers, trans people,queer people, immigrants, medical patients (by their providers),neurodivergent people, and visibly or vocally disabled people.” T

      If accounts can be suspended based on violent behavior such as threats, sexism or perhaps even inappropriate content or posts, what stops AI from being limited on these sites? In the last year a conversation around X's Grok AI supports these issues as many users were requesting inappropriate content from the model, in which Grok delivered via anti-semitic text to graphic sexual photos that violated X's terms of service. The intersection of AI and human bias is dangerous and subject to regulation.

    2. typicallyashj 26 May 2026
      In thecase of US and UK English, this means that white supremacist andmisogynistic, ageist, etc. views are overrepresented in the trainingdata, not only exceeding their prevalence in the general populationbut also setting up models trained on these datasets to furtheramplify biases and harms.

      If AI is a for profit for research pursuit, how would biases benefit any result of AI's output if it appeals to personal opinions rather than factual or impartial results? In terms of scientific research, Artificial Intelligence must uphold factual information and standards to produce quality information and cannot maintain personal biases. I can see how generative AI can possibly connect to a users outlook when conversing or creating works surrounding sensitive topics, but legally would production of biases by AI cause a dispute?

    3. typicallyashj 26 May 2026
      In addition to these calls for documentation and technical fixes,Bietti and Vatanparast underscore the need for social and politicalengagement in shaping a future where data driven systems haveminimal negative impact on the environment [16]

      How would AI companies and their expansion intersect with politicians' and the citizen's vote on this matter? Perhaps through environmental regulations data centers can be limited in size or distance from residential areas, or would water usage be limited on a per day-year scale?

    4. typicallyashj 26 May 2026
      Initiatives such as the SustainNLP workshop5 have since takenup the goal of prioritizing computationally efficient hardware andalgorithms. Schwartz et al. [ 115] also call for the development ofgreen AI, similar to other environmentally friendly scientific de-velopments such as green chemistry or sustainable computing.

      Posing a general question: how does AI energy emissions and resources compare to general energy demands for electricity? Furthermore, would an expansion of "green AI" lead to a greater expansion of renewable or green energy initiatives as a whole, or would the increased output of green energy not even compute with the consumption?

    5. typicallyashj 26 May 2026
      Similar to [ 14 ], we understand the term language model (LM) torefer to systems which are trained on string prediction tasks: that is,predicting the likelihood of a token (character, word or string) giveneither its preceding context or (in bidirectional and masked LMs)its surrounding context.

      The assumption-based structure of "string predication" poses a threat of interpreting information through lenses of common results or probability rather than facts that support it.

    6. typicallyashj 26 May 2026
      As we outline in §3, increasing the environmental andfinancial costs of these models doubly punishes marginalized com-munities that are least likely to benefit from the progress achievedby large LMs and most likely to be harmed by negative environ-mental consequences of its resource consumption.

      As summarized in this passage, these LMs input cost and environmental effects are being forced upon communities that do not agree to the expansion of these models and face real results in their communities such as raised electric bills or degradation of air quality.

    Visit annotations in context

    Annotators

    URL

    markcmarino.com/readings/150/maw/parrots.pdf