Myke Walton

0 followers · 719 views

on the atlas · 36

highlights · 567

AI systems are inherently sociotechnical in na- ture, meaning they are influenced by societal dynamics and human behavior.
DS_AI_Governance_Policy_Brief.pdf
Because real-world uses of AI are always embedded within larger social institutions and power dynamics, technical assessments alone are insufficient to govern AI
DS_AI_Governance_Policy_Brief.pdf
Evaluation is not standard practice in part because our methodology is vague. Where other sciences have standard experimental methods and analytic techniques, we have faith—often groundless and misleading—that building programs is somehow informative
How Evaluation Guides AI Research - Cohen - 1988 - AI Magazine - Wiley Online Library
view the prediction of protein structures as a graph inference problem in 3D space in which the edges of the graph are defined by residues in proximity.
Highly accurate protein structure prediction with AlphaFold
The Evoformer blocks contain a number of attention-based and non-attention-based components. We show evidence in ‘Interpret - ing the neural network’ that a concrete structural hypothesis arises early within the Evoformer blocks and is continuously refined.
Highly accurate protein structure prediction with AlphaFold
computational intractability of molecular simulation, the context dependence of protein stability and the difficulty of producing sufficiently accurate models of protein physics
Highly accurate protein structure prediction with AlphaFold
The development of computational methods to predict three-dimensional (3D) protein structures from the protein sequence has proceeded along two complementary paths that focus on either the physical interactions or the evolutionary history.
Highly accurate protein structure prediction with AlphaFold
Public input processes are necessary but not sufficient for democratization. A democratic AI ecosystem is one that is good for people, not just one that asks them questions at regular intervals
A Roadmap to Democratic AI
Labs and companies should actively commit to a better supply chain for AI (including around data labor, compute, and other inputs). Model development is an industrial process that mirrors and replicates other forms of geographic extraction (through human labor, natural resource consumption, and more.
A Roadmap to Democratic AI
Build open source tools for collective-fine tuning
A Roadmap to Democratic AI
It is easy to assume that democratizing AI is as simple as investing in the open source AI movement, however, open source is not necessarily democratic, and “what people want” can be at odds with the direction of open source AI
A Roadmap to Democratic AI
Work on the open source development of a suite of different reward models to be used for reinforcement learning from human feedback. This will allow us to better understand methods of fine-tuning models and driving forward value and preference alignment
A Roadmap to Democratic AI
Open-source reward models and other methods of representing collective preferences
A Roadmap to Democratic AI
recognizing that the democratization of access without governance rights is not enough to ensure the public interest.
A Roadmap to Democratic AI
We’re far from the best containers in which to build transformative technology.
A Roadmap to Democratic AI
The core material inputs to AI (data and compute) are governed non-monopolistically. Sites of development and deployment, whether open source movements, corporations, startups, or government agencies, are subject to checks and balances to mitigate against power centralization. Impact: We’re not just gathering collective input, we’ve shifted incentives and built institutional capacity to actually compel action based on the public interest
A Roadmap to Democratic AI
This includes responsively building AI systems to target real community needs, and enabling responsive opt-out of AI systems
A Roadmap to Democratic AI
It would be a shame to make the progress envisioned in this proposal and have it undermined by backdoor exceptions.
Opinion | A.I. Use by Law Enforcement Must Be Strictly Regulated - The New York Times
The federal government regularly offers federal funding as a carrot to win compliance from state and local agencies with federal rules. It should do the same here.
Opinion | A.I. Use by Law Enforcement Must Be Strictly Regulated - The New York Times
The F.B.I., the Transportation Security Administration and other federal agencies are aggressively embracing facial recognition and other biometric technologies that can recognize individuals by their unique physical characteristics.
Opinion | A.I. Use by Law Enforcement Must Be Strictly Regulated - The New York Times
Technologies that are clearly shown to be discriminatory should not be used.
Opinion | A.I. Use by Law Enforcement Must Be Strictly Regulated - The New York Times
Given the rapid adoption of these tools, without evidence of equity or efficacy and with insufficient attention to preventing mistakes, we fully anticipate some A.I. technologies will not meet the proposed standards and their use will be banned for noncompliance.
Opinion | A.I. Use by Law Enforcement Must Be Strictly Regulated - The New York Times
It sets out criteria for A.I. technologies that, without safeguards, could put people’s safety or well-being at risk or violate their rights. If these proposed “minimum practices” are not met, technologies that fall short would be prohibited after next Aug. 1.
Opinion | A.I. Use by Law Enforcement Must Be Strictly Regulated - The New York Times
Ms. Woodruff was eight months pregnant when she was falsely accused of carjacking and robbery; Mr. Williams was arrested in front of his wife and two young daughters as he pulled into his driveway from work. Mr. Oliver lost his job as a result.
Opinion | A.I. Use by Law Enforcement Must Be Strictly Regulated - The New York Times
predictable and preventable harms from law enforcement’s use of emerging technologies. These include false arrests and police seizures, including a family held at gunpoint, after people were wrongly accused of crimes because of the irresponsible use of A.I.-driven technologies including facial recognition and automated license plate readers.
Opinion | A.I. Use by Law Enforcement Must Be Strictly Regulated - The New York Times
Foremost among them is a provision that would allow senior officials to seek waivers by arguing that the constraints would hinder law enforcement. Those law enforcement agencies should instead be required to provide verifiable evidence that A.I. tools they or their vendors use will not cause harm, worsen discrimination or violate people’s rights.
Opinion | A.I. Use by Law Enforcement Must Be Strictly Regulated - The New York Times
With the new executive order and its companion OMB memo, civil society organizations should seize the opportunity to voice concerns, submit regulatory comments (including to the OMB), and collaborate with agencies to shape governance standards.
I Implemented a Federal Government Executive Order on Technology in Mexico. Here’s What I Learned | by Ania Calderon | Data & Society: Points | Nov, 2023 | Medium
Tech skills alone are not enough: The government must create interdisciplinary spaces where people from technical, social, and political backgrounds are able to ask probing questions and seek the perspectives of people with diverse lived experiences.
I Implemented a Federal Government Executive Order on Technology in Mexico. Here’s What I Learned | by Ania Calderon | Data & Society: Points | Nov, 2023 | Medium
Having core competencies and timely access to infrastructure within government can help mitigate the risk of corporate capture.
I Implemented a Federal Government Executive Order on Technology in Mexico. Here’s What I Learned | by Ania Calderon | Data & Society: Points | Nov, 2023 | Medium
We offered training across all federal government agencies and saw the best outcomes in departments where we seconded multidisciplinary units to work in partnership with agencies that had political acumen but lacked technical skills.
I Implemented a Federal Government Executive Order on Technology in Mexico. Here’s What I Learned | by Ania Calderon | Data & Society: Points | Nov, 2023 | Medium
While we worked to implement a mandate to improve the governance and availability of government-held data, we focused on socializing the tangible benefits agencies would see if they committed to this kind of change, rather than relying on the logic of compliance.
I Implemented a Federal Government Executive Order on Technology in Mexico. Here’s What I Learned | by Ania Calderon | Data & Society: Points | Nov, 2023 | Medium
To implement a strategy, people need the right leadership and support.
I Implemented a Federal Government Executive Order on Technology in Mexico. Here’s What I Learned | by Ania Calderon | Data & Society: Points | Nov, 2023 | Medium
As Gilman outlines, public participation helps avert harmful impacts, adds legitimacy to decisions, and improves accountability
I Implemented a Federal Government Executive Order on Technology in Mexico. Here’s What I Learned | by Ania Calderon | Data & Society: Points | Nov, 2023 | Medium
Nurturing trusted relationships with community groups and civil society organizations, and establishing meaningful participation processes with the people you serve, should be baked into every agency’s strategy.
I Implemented a Federal Government Executive Order on Technology in Mexico. Here’s What I Learned | by Ania Calderon | Data & Society: Points | Nov, 2023 | Medium
Lasting change has a better chance when the moment of top-down authority is met with on the ground legitimacy — especially when political winds shift.
I Implemented a Federal Government Executive Order on Technology in Mexico. Here’s What I Learned | by Ania Calderon | Data & Society: Points | Nov, 2023 | Medium
Now the hard work begins, to build a robust public debate about what we want AI systems to do and where there is a need for guardrails to mitigate their harms.
I Implemented a Federal Government Executive Order on Technology in Mexico. Here’s What I Learned | by Ania Calderon | Data & Society: Points | Nov, 2023 | Medium
It was this “science of delivery” that I had to learn quickly, by understanding how to navigate the structures in place and use existing vehicles creatively to advance policy goals.
I Implemented a Federal Government Executive Order on Technology in Mexico. Here’s What I Learned | by Ania Calderon | Data & Society: Points | Nov, 2023 | Medium
this means striking a skillful balance between policy design and implementation, and making the structures you have work for the benefit of people’s lives in tangible and specific ways.
I Implemented a Federal Government Executive Order on Technology in Mexico. Here’s What I Learned | by Ania Calderon | Data & Society: Points | Nov, 2023 | Medium
it is important to ensure that their sense of urgency does not undercut the intensive and often slow-moving work that sustainable change requires.
I Implemented a Federal Government Executive Order on Technology in Mexico. Here’s What I Learned | by Ania Calderon | Data & Society: Points | Nov, 2023 | Medium
A convivial techno-social system is one that can come into balance with its environment, where recursive coordination starts with autonomy from below.
Predicted benefits, proven harms
Researchers and social movements can’t restrict themselves to reacting to the machinery of AI but need to generate social laboratories that prefigure alternative socio-technical forms
Predicted benefits, proven harms
A movement to replace AI will call into question both material and conceptual boundaries and will mobilise care and mutual aid in the face of abstractions that enact actual violence
Predicted benefits, proven harms
One way to counter the risk of venture capitalists turning AI into a weapon against the rest of us would be to put care at the centre of both research and action
Predicted benefits, proven harms
AI concentrates and condenses these microfascisms through the way it enacts states of exception, transformative violence and essentialised othering.
Predicted benefits, proven harms
They are also bullshit engines in the social sense because, like marketing and middle management, their apparently reasoned arguments are actually completely detached from lived experience.
Predicted benefits, proven harms
They are bullshit engines in the technical sense because they optimise on the generation of plausible natural language text based not on a model of causal relations and real meanings but blindly, based on preprogrammed statistical patterns in a vast (but never full) corpus of existing language.
Predicted benefits, proven harms
AI is not a way of representing the world but an intervention that helps to produce the world that it claims to represent. Setting it up one way or another changes what becomes naturalised and what becomes problematised. Who gets to set up the AI becomes a crucial question of power.
Predicted benefits, proven harms
Instead of sci-fi futures, what we get is the return of 19th-century industrial relations and the dissolution of post-war social contracts.
Predicted benefits, proven harms
The compulsion to show balance by always referring to AI’s alleged potential for good should be dropped; we must acknowledge that the social benefits are still speculative, but the harms have been empirically demonstrated
Predicted benefits, proven harms
AI increases what historian and philosopher Hannah Arendt called “institutional thoughtlessness” – the inability to critique instructions or reflect on consequences. Its objective devaluations interface all too readily with existing bureaucratic cruelties, scaling administrative violence in ways that intensify structures of inequality
Predicted benefits, proven harms