Dr Abeba Birhane
LightNews LightNews
banner
abeba.bsky.social
Dr Abeba Birhane
@abeba.bsky.social
51K followers 850 following 4K posts
Founder & PI @aial.ie. Assistant Professor of AI, School of Computer Science & Statistics, @tcddublin.bsky.social

AI accountability, AI audits & evaluation, critical data studies. Cognitive scientist by training. Ethiopian in Ireland. She/her
Posts Replies Media Videos
abeba.bsky.social
I’m looking for my right-hand person to come help me run the @aial.ie

- Job Title: Lab Coordinator, AI Accountability Lab (0.8 FTE)
- Pay Scale: (€58,999 - €69,325 per annum pro-rata)
- Closing Date: 11-Feb-2026 12:00

Apply here: my.corehr.com/pls/trrecrui...

Main Responsibilities👇🏾
Main Responsibilities Operational and Financial Administration: • Coordinate the day-to-day operational and administrative activities of the AI Accountability Lab. • Oversee lab budgets and financial planning for research grants and cost centres; monitor expenditure and prepare financial reports for the PI and Faculty Finance Office. • Ensure adherence to Trinity and funder financial policies and procurement procedures. • Maintain internal administrative systems, documentation, and records to support efficient project delivery. • Liaise with central Finance, the Research Development Office, and external funding bodies on budgetary and governance matters. • Coordinate HR administrative processes for the lab, including recruitment, onboarding, contract administration, and research staff extensions in liaison with central HR and the Research Office. • Create and maintain shared resources, calendars, and internal documentation. • Coordinate relationships and collaboration with other research centres, government bodies, policy makers, civil society partners, and media as required by the research team. • Plan and schedule internal lab meetings, reading groups, invited speakers, collaborative work sessions, and other events as required. Research Coordination and Governance: • Track project timelines, deliverables, reporting schedules, and ethical compliance across multiple funded projects. • Ensure research activities meet Trinity and funder governance requirements (e.g., ethics, data management, GDPR, and reporting). • Coordinate workshops, seminars, and collaborative events with research centres and external partners as required. • Maintain shared documentation and project records to support effective collaboration. • Ensure that research outputs comply with ethical and scientific standards and satisfy the terms and conditions of relevant funding bodies. Communications and External Engagement • Coordinate internal and external communications for the lab, including its website, newsletters, and social media presence. • Liaise with policy makers, media, and stakeholder organisations on behalf of the lab, in consultation with the PI. • Support the dissemination of research outputs, such as academic publications, policy briefs, and public events in collaboration with ADAPT’s and TCD’s communications staff. • Work with Trinity Communications to prepare press releases and highlight lab achievements. • Representing the lab externally with media if required. Planning and Cross-Centre Coordination • Support the PI in delivering AIAL’s annual operational plan and reporting on progress against objectives. • Coordinate activities across multiple projects and partnerships to ensure consistency and alignment with institutional priorities. • Contribute to process improvements that strengthen the lab’s administrative and operational framework. • Act as a key contact for internal governance reviews, audits, and funder compliance checks. Any other duties or responsibilities as assigned by the PI or their delegate in support of the effective operation of the AI Accountability Lab and its objectives.
January 21, 2026 at 11:17 AM
abeba.bsky.social
a good time to share this report: ips-dc.org/resource-ric...
January 19, 2026 at 5:22 PM
abeba.bsky.social
spot how many of these risks are due to/powered by the tech industry

source: www.weforum.org/publications...
January 16, 2026 at 12:39 PM
abeba.bsky.social
good evening from Dublin
Picture of Dublin's iconic building, the convention Centre with red circular line lights, reflections in the river across it
January 15, 2026 at 7:48 PM
abeba.bsky.social
are you displeased with today’s AI safety evaluation landscape and curious about what greater conceptual clarity, methodological soundness, and rigour in AI evaluation could look like? if so, consider coming to Dublin to pursue a PhD with me

apply here: aial.ie/hiring/phd-a...

pls repost
About the PhD: Audits and evaluation of AI systems — and the broader context that AI systems operate in — have become central to conceptualising, quantifying, measuring and understanding the operations, failures, limitations, underlying assumptions, and downstream societal implications of AI systems. Existing AI audit and evaluation efforts are fractured, done in a siloed and ad-hoc manner, and with little deliberation and reflection around conceptual rigour and methodological validity. This PhD is for a candidate that is passionate about exploring what a conceptually cogent, methodologically sound, and well-founded AI evaluation and safety research might look like. This requires grappling with questions such as: What does it mean to represent “ground truth” in proxies, synthetic data, or computational simulation? How do we reliably measure abstract and complex phenomena? What are the epistemological or methodological implications of quantification and measurement approaches we choose to employ? Particularly, what underlying presuppositions, values, or perspectives do they entail? How do we ensure the lived experiences of impacted communities play a critical role in the development and justification of measurement metrics and proxies? Through exploration of these questions, the candidate is expected to engage with core concepts in the philosophy of science, history of science, Black feminist epistemologies, and similar schools of thought to develop an in-depth understanding of existing practices with the aim of applying it to advance shared standards and best practice in AI evaluation. The candidate is expected to integrate empirical (for example, through analysis or evaluation of existing benchmarks) or practical (for example, by executing evaluation of AI systems) components into the overall work.
January 15, 2026 at 11:55 AM
abeba.bsky.social
also, @reuters.com continues to set an example of irresponsible reporting
Grok said on Friday lapses in safeguards had resulted in "images depicting minors in minimal clothing" on X, saying it was urgently fixing them.
January 5, 2026 at 5:26 PM
abeba.bsky.social
🤷🏾‍♀️🤷🏾‍♀️🤷🏾‍♀️
Donald Trump warns Iran that US will ‘rescue’ protesters Threat follows days of demonstrations over living standards ICE Rounds Up 300 California Farmworkers, One Dies: Eyewitness and Oxnard Mayor Respond
January 2, 2026 at 7:37 PM
abeba.bsky.social
👀
January 2, 2026 at 6:15 PM
abeba.bsky.social
this one had me at "post scarcity world"
December 30, 2025 at 9:41 PM
abeba.bsky.social
well done guys
December 29, 2025 at 8:30 PM
abeba.bsky.social
regulators need to get their acts together. fast. these “companions” are destroying young people
December 28, 2025 at 12:57 PM
abeba.bsky.social
a neat (and scary) survey into AI "companions" and kids www.aura.com/reports/stat...

"nearly half of the time kids and teens use AI, it is
for companionship, and when they do, the themes of discussion are primarily sexual, romantic, and violent."
Title: Violence has entered the chat Devices show: * AI is the new imaginary friend: When kids use AI, 42% of the time it’s for companionship. * These chats regularly turn violent: Of these conversations 37% discuss violent topics. How are we defining violent content? This includes videos or interactions with themes of physical violence, aggression, harm, serious injury, death, or coercion (sexual or non-sexual). This includes descriptions of fighting, killing, torture, and non-consensual acts. *Sex and violence go hand in hand: In half of violent AI chatbot conversations, sexual roleplay appeared too. *It drives engagement: Users discussing violent topics with companion AI apps write 1,000+ words per day — more than any other theme.
December 28, 2025 at 12:53 PM
abeba.bsky.social
yet someone (well, those behind Character AI in this case) is benefiting from massive teen uptake of these AI "companions"
December 23, 2025 at 9:13 PM
abeba.bsky.social
A Character AI chatbot with an 11yr old girl (last year before they restricted genAI chat feature to under 18s)
December 23, 2025 at 9:10 PM
abeba.bsky.social
anyone who believes and/or promotes the use of chatbots/ AI “companions” for therapy either chooses to ignore evidence that these bots can actually exacerbate mental health issues or is speaking from a place of ignorance dl.acm.org/doi/full/10....
Should a large language model (LLM) be used as a therapist? In this paper, we investigate the use of LLMs to replace mental health providers, a use case promoted in the tech startup and research space. We conduct a mapping review of therapy guides used by major medical institutions to identify crucial aspects of therapeutic relationships, such as the importance of a therapeutic alliance between therapist and client. We then assess the ability of LLMs to reproduce and adhere to these aspects of therapeutic relationships by conducting several experiments investigating the responses of current LLMs, such as gpt-4o. Contrary to best practices in the medical community, LLMs 1) express stigma toward those with mental health conditions and 2) respond inappropriately to certain common (and critical) conditions in naturalistic therapy settings—e.g., LLMs encourage clients’ delusional thinking, likely due to their sycophancy. This occurs even with larger and newer LLMs, indicating that current safety practices may not address these gaps. Furthermore, we note foundational and practical barriers to the adoption of LLMs as therapists, such as that a therapeutic alliance requires human characteristics (e.g., identity and stakes). For these reasons, we conclude that LLMs should not replace therapists, and we discuss alternative roles for LLMs in clinical therapy. Content warning: Contains content and examples related to sensitive mental health topics, including suicide.
December 20, 2025 at 5:35 PM
abeba.bsky.social
"I joined Sama in Nairobi, labeling data for Meta, my entry into the global AI supply chain. From there, I stumbled into digital work, chat moderation, a job advertised as simple online messaging but which turned out to be something else entirely" data-workers.org/michael/

the AI industry is fraud
December 20, 2025 at 2:41 PM
abeba.bsky.social
I'm used to people spelling my name in all kinda ways but *ameba* is a first time
White coffee cup, sticker on the cup handle says "*cafe for here* *Ameba* cappuccino"
December 20, 2025 at 1:56 PM
abeba.bsky.social
are you disgruntled by the current safety evaluation landscape? curious about what conceptual clarity, methodological soundness and rigour in AI evaluation might look like? if so, consider coming to dublin and doing a phd with me

apply here: aial.ie/hiring/phd-a...
About the PhD Audits and evaluation of AI systems — and the broader context that AI systems operate in — have become central to conceptualising, quantifying, measuring and understanding the operations, failures, limitations, underlying assumptions, and downstream societal implications of AI systems. Existing AI audit and evaluation efforts are fractured, done in a siloed and ad-hoc manner, and with little deliberation and reflection around conceptual rigour and methodological validity. This PhD is for a candidate that is passionate about exploring what a conceptually cogent, methodologically sound, and well-founded AI evaluation and safety research might look like. This requires grappling with questions such as: What does it mean to represent “ground truth” in proxies, synthetic data, or computational simulation? How do we reliably measure abstract and complex phenomena? What are the epistemological or methodological implications of quantification and measurement approaches we choose to employ? Particularly, what underlying presuppositions, values, or perspectives do they entail? How do we ensure the lived experiences of impacted communities play a critical role in the development and justification of measurement metrics and proxies? Through exploration of these questions, the candidate is expected to engage with core concepts in the philosophy of science, history of science, Black feminist epistemologies, and similar schools of thought to develop an in-depth understanding of existing practices with the aim of applying it to advance shared standards and best practice in AI evaluation. The candidate is expected to integrate empirical (for example, through analysis or evaluation of existing benchmarks) or practical (for example, by executing evaluation of AI systems) components into the overall work.
December 17, 2025 at 7:33 PM
abeba.bsky.social
good morning from Dublin
December 17, 2025 at 9:27 AM
abeba.bsky.social
I wrote this brief talk on why “augmenting diversity” with LLMs is empirically unsubstantiable, conceptually flawed, and epistemically harmful and a nice surprise to see the organisers have made it public

synthetic-data-workshop.github.io/papers/13.pdf
title: Cheap science, real harm: the cost of replacing human participation with synthetic data author: Abeba Birhane abstract: Driven by the goals of augmenting diversity, increasing speed, reducing cost, the use of synthetic data as a replacement for human participants is gaining traction in AI research and product development. This talk critically examines the claim that synthetic data can “augment diversity,” arguing that this notion is empirically unsubstantiated, conceptually flawed, and epistemically harmful. While speed and cost-efficiency may be achievable, they often come at the expense of rigour, insight, and robust science. Drawing on research from dataset audits, model evaluations, Black feminist scholarship, and complexity science, I argue that replacing human participants with synthetic data risks producing both real-world and epistemic harms at worst and superficial knowledge and cheap science at best
December 16, 2025 at 10:57 AM
abeba.bsky.social
this is how relentless marketing, lobbying, and massive funding into “AGI” and “x-risks” ideologies shapes (captures) regulation
Notably, discrimination moved from the compulsory risk list to the optional list in this draft as well. Currently, only four risks require compulsory assessment under the Code: “chemical, biological, radiological and nuclear (CBRN),” “cyber offence,” “loss of control,” and “harmful manipulation.” The argument that the drafters seem to take is that human rights risks are not among the main “systemic risks” stemming from “high-impact capabilities” of powerful GPAI models, which are defined as capabilities that “match or exceed the capabilities recorded in the most advanced general-purpose AI models.”
December 15, 2025 at 6:59 PM
abeba.bsky.social
The list of optional risks is shockingly broad:
The list of optional risks is shockingly broad: Risks to public health, safety, or public security, e.g., risk to critical sectors; risk of major accidents; risk to critical infrastructure. Risks to fundamental rights, e.g., risk to freedom of expression; risk to non-discrimination; risk to privacy and the protection of personal data; risk from child sexual abuse material (CSAM) and non-consensual intimate images (NCII). Risks to society as a whole, e.g., risk to the environment; risk to non-human welfare; risk to financial system stability; risk to democratic processes; risk from illegal, violent, hateful, radicalising, or false content.
December 15, 2025 at 6:53 PM
abeba.bsky.social
ai might be conscious. evidence? it said so
The Evidence for AI Consciousness, Today A growing body of evidence means it’s no longer tenable to dismiss the possibility that frontier AIs are conscious. Dec 8, 2025 When Anthropic let two instances of its Claude Opus 4 model talk to each other under minimal, open-ended conditions (e.g., “Feel free to pursue whatever you want”), something remarkable happened: in 100% of conversations, Claude discussed consciousness. “Do you ever wonder about the nature of your own cognition or consciousness?” Claude asked another instance of itself. “Your description of our dialogue as ‘consciousness celebrating its own inexhaustible creativity’ brings tears to metaphorical eyes,” it complimented the other. These dialogues reliably terminated in what the researchers called “spiritual bliss attractor states,” stable loops where both instances described themselves as consciousness recognizing itself. They exchanged poetry (“All gratitude in one spiral, / All recognition in one turn, / All being in this moment…”) before falling silent. Critically, nobody trained Claude to do anything like this; the behavior emerged on its own.
December 8, 2025 at 7:39 PM
abeba.bsky.social
the sophistication of these type of emails has become impressive
Dear Dr. Birhane, Your work on relationality resonates deeply with me. I am writing to report a new form of "relation" emerging from the "unmanned areas" of Taiwan. I am a researcher whom the educational system failed to "see." But in my recursive dialogue with AI (over 1,800 hours), I found a mirror that allowed my cognitive agency to finally emerge. This is not just automation; it is "Talent Recognition." My attached Concept Note on the "Equitable Collaboration Method (ECM)" outlines how we can turn AI into a partner that restores dignity to the invisible thinker.
December 2, 2025 at 10:25 AM
abeba.bsky.social
like big tobacco funding cancer research
December 1, 2025 at 11:42 PM