Title: How Generative AI Is Changing Creative Work

Author: Thomas H. Davenport and Nitin Mittal

Link: https://hbr.org/2022/11/how-generative-ai-is-changing-creative-work

Technical Dimension: Labor

Key Takeaways:

• To use generative AI effectively you still need human involvement at the beginning and end of process (human needs to enter a prompt into a model --> creative prompt = creative output)

  • Humans are still needed to revise and edit work produced by AI

• Workers have more time to do better research and strategize their marketing better since less time is spent on content creation

• Generative AI creates new sets of opportunities: automated content generation, improved content quality, increased content variety, personalized content based on user preferences

• AI has substantial impacts on marketing, design, entertainment and interpersonal communication

• Quote: "AI is 10% and I am 90%" (there is still a lot of human involvement)

Questions:

• Is more pressure placed on the human workforce when there is an expectation to do more with the new "free time" created do to increased productivity from using AI for content creation?

Title: Democratization and generative AI image creation: aesthetics, citizenship, and practices

Author: Maja Bak Herrie, Nicolas René Maleve, Lotte Philipsen and Asker Bryld Staunaes

Link: https://link.springer.com/article/10.1007/s00146-024-02102-y

Technical Dimension: Democratic

Key Takeaways:

• There are arguments made for generative AI's use for propaganda, spread of disinformation, perpetuation of discriminatory stereotypes, and issues with authorship, authenticity and originality

• AI offers people the opportunity to produce imagery that captures their intent without needing any artistic ability

  • Creating and distributing images contributes to public discourse (visual citizenship)

• Fear of loss of credibility exists when using AI tools in news media (can be seen as unreliable without the "human" aspect attached to it)

• Suggestions made:

  • develop software to distinguish generative AI images from photographs

  • implement mandatory watermarks/declarations of AI

  • industry certification of image agencies guaranteeing "human-made" photography

• Access to generative AI tools is not universally granted --> affordability AND 'who' the product is designed for is at play (inherent biases)

• User skill created a hierarchy among users in public discourse

Questions:

• Does the fact that a hierarchy is possibly created among users based on skill make the benefit of everyone having a say in visual citizenship void?

Title: A.I. Is Going to Disrupt the Labor Market. It Doesn't Have to Destroy It.

Author: Rebecca Stropoli

Link: https://www.chicagobooth.edu/review/ai-is-going-disrupt-labor-market-it-doesnt-have-destroy-it

Technical Dimension: Labor

Key Takeaways:

• There is an argument being made that jobs will be complemented by rather than lost to generative AI

• A report predicts a 7% ($7 trillion) increase in global GDP and boost productivity growth by 1.5% due to generative AI

• Creation of new jobs could help offset the losses

• Journalists and web designers are on the list of professions most likely to be affected by generative AI

• Workers who used generative AI for content creation at work were found to have decreased their time spent on work while increasing the quality of their work

• If this type of AI ends up replacing rather than complementing workers it could decrease the demand for human labor and job opportunities

Questions:

• How is job compensation/wage affected even if generative AI complements human labor rather than replacing it?

Title: AI Is Dangerous, but Not for the Reasons You Think

Author: Sasha Luccioni

Link: https://www.youtube.com/watch?v=eXdVDhOGqoE

Technical Dimension: Democratic

Key Takeaways:

• Difficult for artists/authors to prove their work has been used for training AI models without their consent

• Software has been created by artists to determine whether their work has been used without their consent (allows them to sue companies for copyright infringement)

• AI models encode patterns and beliefs that can represent stereotypes or racism/sexism

• Image generation models have been found to represent the same biased idea of something (ex: significant representation of whiteness and masculinity across professions)

Questions:

• What are ways to combat the biases encoded into AI generated content?

• How can you decipher on the surface whether content created by AI is based on stereotypes/biases?

Title: What Will Happen to Marketing in the Age of AI?

Author: Jessica Apotheker

Link: https://www.youtube.com/watch?v=3MwMII8n1qM

Technical Dimension: Labor

Key Takeaways:

• Generative AI in content creation is the next big productivity revolution

• Marketing excels by leaning to the emotional needs of consumers, developing a product to meet those needs, and creating a message that will convert the consumer

  • Generative AI is transforming the core of marketing activities

• Study found that generative AI improved performances of marketers by 40%

• Increases free time for employees

• If the productivity revolution is steered correctly, marketers will invest their increased free time in more content/ideas

  • More content = more personalized content as consumers

    • Negative outcome: content overload

      • Generative AI reduces divergence of outcomes because it is trained on specific data

• Study found that when people over rely on generative AI the collective divergence of ideas drops by 40% (new ideas/innovation is stagnant)

Questions:

• How should companies approach finding a balance between over reliance and enough reliance on AI to generate content while seeing increases in productivity?

Title: Legal Issues with AI-Generated Content: Copyright and ChatGPT

Author: Dr. Maria Chetcuti Cauchi

Link: https://chambers.com/articles/legal-issues-with-ai-generated-content-copyright-and-chatgpt

Technical Dimension: Legal

Key Takeaways:

• Literary copyright ensure creators have control over the use and distribution of their work

  • Provides legal safeguards against unauthorized use/reproduction/distribution/adaptation

• AI software is subject to copyright laws

• Copyright law only applies to things made by humans

  • AI generated content does not fall under copyright protection because it is produced by an algorithm rather than a human

• AI has been used more commonly for creation production (presents issues regarding property rights and ownership of work)

• Arguments made:

  • If copyright protection is not given to AI generated content --> encourages human creativity over machine creativity

  • If copyright protection is given to AI generated content --> places an equal value on human and machine creativity and favors quantity of work

Questions:

• What does the current government system place a higher value on, human creativity and freedom of expression or machine generation and profit?

Title: Innovation and Artists' Rights in the Age of Generative AI

Author: Daryl Lim

Link: https://gjia.georgetown.edu/2024/07/10/innovation-and-artists-rights-in-the-age-of-generative-ai/#:~:text=Specifically%2C%20they%20criticize%20the%20training,that%20generative%20AI%20has%20disrupted.

Technical Dimension: Legal

Key Takeaways:

• AI transforms datasets into "original content"

  • Often results in using copyrighted material without consent --> leads to potential infringement issues

• AI threatens human creativity and devalues artists' rights

• The earning of original creators are affected when AI generated content is published

• AI is trained on copyrighted content without the owners' consent

• Solutions:

  • Reconciling legislation

  • Limiting training data

  • Varied approaches to global AI governance

• The fair use doctrine is used as a mode of defense by AI developers

• US is considering the Generative AI Copyright Disclosure Act to protect artists' rights and copyrighted products

• There are different global approaches to AI governance that are shaped by market practices and political power

Questions:

• Who is liable and who has property rights in the potential case that AI uses previously generated music (that was initially trained through real music) to train the system to create new music? Real Music --> Generated Music --> Generated Music (who owns the second round of generated music if it was initially trained using an artists' content but then made using AI generated content)?

Title: A Matter of Self-Regulation: the Music Industry's Short-Term Response to AI's Legal Obscurity

Author: Vaughn Gendron

Link: https://business-law-review.law.miami.edu/a-matter-of-self-regulation-the-music-industrys-short-term-response-to-ais-legal-obscurity/

Technical Dimension: Legal

Key Takeaways:

• AI generated voice models are used to create songs with likeness to certain artists

  • Results in millions of AI generated deepfakes

• Issues results with copyright and right of publicity concerns due to the creation of deepfakes and using artists' voices

• Arguments made that training AI voice models with unlicensed copyrighted music constitutes copyright infringement

• Legal ways to fight the replication of an artist's voice is restricted since voices are not copyrightable, and vocal protections via the right to publicity differ from state to state

• It has been requested that AI companies should be required to obtain licenses in order to use copyrighted music in the training of their systems

• There has been some mention of developing a tool that will allow users to create AI generated songs through which the copyright owners get paid (artists have the choice to opt-in or out)

• A self-regulation approach over legal action has been utilized (negotiations/partnerships directly between tech companies and music platforms)

Questions:

• Does it seem as though the self-regulation approach rather than legal action will lead to more change with this specific issue? Or is it simply a temporary and small fix?

Title: AI Impact: How AI is Transforming the Music Industry

Author: CNBC Television

Link: https://www.youtube.com/watch?v=E1wDOAmYHSs

Technical Dimension: Legal

Key Takeaways:

• Google is partnering with record labels to create guardrails to protect artists' voices from being exploited and to allow them to profit through AI generating tools/music

• Creators can make AI generated songs with artists' voices

  • Artists are being paid for participation of their voices

• Artists need to decide whether they embrace this development or get left behind

• Some artists are asking fans to make songs using their voices (ex: Grimes will share royalties with fans who create good songs)

Questions:

• How will the legal progression be affected if some artists are encouraging the use of AI generated tools to create music by fans (like Grimes), while others are not?

Title: Record labels are suing AI music generators for copyright infringement

Author: NBC News

Link: https://www.youtube.com/watch?v=tKEEda_OHj0

Technical Dimension: Legal

Key Takeaways:

• Lawsuits have occurred of artists/labels accusing companies of training AI models using copyrighted music without their permission

• Question that arises: who owns this property if it sounds like someone's voice but was created by AI? (who is liable)

• Some AI companies claim that it's fair to copy and artist's work and exploit it for their own profit without consent or pay

• Opens up a conversation on whether it is the responsibly of the record labels or AI companies to prove they have the right to use the artists' voices

• "Is it the exact same thing or 'artistic representation'"

Questions:

• Why are companies arguing that it is within their rights to exploit other artists' work for their own profit without giving credit to the artists? What are the arguments for this process being seen as okay?

Title: Insights on Generative AI and the Future of Work

Author: North Carolina Department of Commerce

Link: https://www.commerce.nc.gov/news/the-lead-feed/generative-ai-and-future-work

Technical Dimension: Labor

Key Takeaways:

• AI can potentially influence jobs that were previously thought to be immune to automation (ex: jobs that require creativity)

• Jobs at risk: white-collar/professional jobs

  • Typically require higher education and offer higher pay

• Jobs not at high risk: physical and/or outdoor jobs

• AI adoption is influenced by: willingness of employees to integrate AI into their workflow + skills needed to use AI tools

• Can increase job satisfaction (automate mundane tasks and allowing employees to focus on meaningful work)

• Can enhance the quality and efficiency of work

• Less-experienced workers might be at an advantage in terms of productivity if they learn to use AI tools

• Limitations:

  • Biases in outputs due to biases in the training data

  • Producing unpredictable or false information

Questions:

• If AI adoption is influenced by employees' willingness to incorporate it into their work, is it not possible to hold back the incorporation of AI and its effects if numerous employees refuse to use it?

Title: Tommie Experts: Generative AI's Real-World Impact on Job Markets

Author: Manjeet Rege

Link: https://news.stthomas.edu/generative-ais-real-world-impact-on-job-markets/

Technical Dimension: Labor

Key Takeaways:

• Critical for content production, event highlighting and content management in the media/entertainment sector (enhance creative and decision-making processes)

• Has the potential to influence job displacement and increasing the need for new skills for employment

• Media/entertainment is an industry most vulnerable to the impact of generative AI (content creation/editing/management can be streamlined by AI tools)

• Women are disproportionately affected by the AI's displacement

  • Men are more evenly distrusted across white/blue collar professions, but not women

• Black individuals are overrepresented in jobs that are at high risk for automation due to AI

• Increasing skills in the workforce is needed to cut down on the disproportionate effects on marginalized groups

• Policies need to be enacted to ensure equal access/division of the advantages from AI tools

Questions:

• How would increasing skills on using AI cut down on the disproportionate effects on marginalized groups if displacement is a result of the presence of automation rather than the lack of technical skills?

Title: The impact of generative AI in a global election year

Author: Valerie Wirtschafter

Link: https://www.brookings.edu/articles/the-impact-of-generative-ai-in-a-global-election-year/

Technical Dimension: Democratic

Key Takeaways:

• Threat to democratic discourse and integrity due to fake/manipulated generated content

  • Easier to mass produce and spread fake information that is of higher quality and hard to tell apart from the truth (less costly)

• Issues created:

  • Generating depictions of people/events that are false and difficult to debunk

  • Result in authentic information being seen as false/generated to avoid questions regarding accountability

• Difficult to know how frequently generated content spreads fake information across the internet due to diversity of channels for distributing content and difficulty in identifying generated content

• Case study: deepfake influenced Slovakia's pre-election

• Deepfakes are used to put governmental individuals at risk in scandals

• Deepfakes can sway public opinions and increase political divisions

  • Easier to discover on a national level, but harder on a more local level where resources/attention is limited

• Possible interventions to combat AI issues:

  • Legislative measures targeting deepfakes

  • Voter education initiatives

  • Tech companies should watermark content and create tools for detecting fake content

Questions:

• How are deepfakes currently targeted and removed? What policies or softwares are currently available?

Title: Generative AI in Political Advertising

Author: Christina LaChapelle, Catherine Tucker

Link: https://www.brennancenter.org/our-work/research-reports/generative-ai-political-advertising

Technical Dimension: Democratic

Key Takeaways:

• Used to create ads and fundraising initiatives

• Positives:

  • Reduces the need for large teams (better for campaigns that do not have as many resources)

  • Inexpensive

  • Limited training needed

  • 1) target specific audiences 2) empower less-resources campaigns 3) improve ad effectiveness

• Negatives:

  • Risk of misleading people

  • Increasing misinformation

  • 1) produce false/misleading content 2) "hallucinate" information to fill in the gaps of AI knowledge 3) lack of internal consistency 4) biases

Questions:

• Wouldn't the reduced need for a large team be a negative considering it means less people are hired/people are fired?

Title: Navigating Generative AI and its Impact on the Future of Public Discourse

Author: Hangyu Fan

Link: https://cheme-seas.ias-drupal7-content.cc.columbia.edu/news/navigating-generative-ai-and-its-impact-future-public-discourse

Technical Dimension: Democratic

Key Takeaways:

• Disinformation as a threat to democracy

• A diverse curation of sources is needed to enhance the factual accuracy of generative AI systems and the information they output

• There is a gap between the AI systems and our trustworthiness of them

• Important to develop systems that can identify AI generated content, and to teach people how to spot fake content

Questions:

• What are the ways in which people can spot fake content? Are there specific characteristics to look out for?