Upcoming Events

Past Events

14 March: Early Career Conference 2025

focused-students-working-on-advanced-robotics

Focused Students Working on Advanced Robotics

Robotics and Embodied Intelligence Workshop

<h1>Robotics and Embodied Intelligence Workshop</h1>
The relationship between the body and mind of a robot has been a topic of debate for centuries, driving scientists to explore intelligent and adaptive behaviour in both animals and machines. In robotics, this relationship is critical to designing systems that can interact meaningfully with their environments, adapt to new challenges, and exhibit humanlike intelligence. Embodied Intelligence studies how physical interactions shape thoughts, emotions, and actions—not just in humans but also in robots.
In the age of AI and Machine Learning, this research is vital, offering insights that enhance traditional AI technologies by integrating physical embodiment and environmental interaction. Robotics plays a central role in this exploration, serving as both a tool for studying intelligence and a field that benefits directly from embodied principles.
This topic is particularly relevant and timely given the underrepresentation of women in robotics. Promoting diverse perspectives is essential for fostering innovation and ensuring that the field reflects a broad range of ideas and experiences. By engaging young researchers, especially women, this workshop aims to inspire participation and contributions to a field where diversity is needed to address complex challenges and drive meaningful progress.
This event is organised by the <a href="https://www.eng.cam.ac.uk/news/cambridge-researchers-showcase-robots-globalwiin">Women-in-Robotics Cambridge</a> initiative and sponsored by the Centre for Human-Inspired AI (CHIA).
<h2>Schedule</h2>
<table>
<tbody>
<tr>
<td>9.45-10:00</td>
<td>Welcome and coffee</td>
</tr>
<tr>
<td>10:00-10:10</td>
<td>Introduction</td>
</tr>
<tr>
<td>10:10-10.30</td>
<td>Yiannis Demiris</td>
</tr>
<tr>
<td>10:30- 10:50</td>
<td>Hatice Gunes</td>
</tr>
<tr>
<td>10:50-11:10</td>
<td>Tom Erez</td>
</tr>
<tr>
<td>11:10-11:40</td>
<td>Panel Discussion</td>
</tr>
<tr>
<td>11:40-12:30</td>
<td>Interactive discussions in focused groups</td>
</tr>
<tr>
<td>12:30-13:30</td>
<td>Lunch</td>
</tr>
<tr>
<td>13:30-13:50</td>
<td>Thrishantha Nanayakkara</td>
</tr>
<tr>
<td>13:50-14:10</td>
<td>Kaspar Althoefer</td>
</tr>
<tr>
<td>14:10-14:40</td>
<td>Panel Discussion</td>
</tr>
<tr>
<td>14:40-14:50</td>
<td>Concluding remarks</td>
</tr>
<tr>
<td>14:50-15:30</td>
<td>Coffee and networking</td>
</tr>
<tr>
<td>15.30-16.30</td>
<td>Bio-inspired robotics lab (BIRL) tour</td>
</tr>
</tbody>
</table>
&nbsp;
Date: Tuesday, 18 February 2025
Location: Lucia Windsor Room, Newnham College, Cambridge
<a href="https://www.eventbrite.co.uk/e/robotics-and-embodied-intelligence-workshop-tickets-1205393025989?aff=oddtdtcreator">Tickets available here</a>

Speakers

max-665-1

Max-665-1

Prof. Max Kleiman-Weiner: Cooperative Minds and Machines

<h1>Prof. Max Kleiman-Weiner: Cooperative Minds and Machines</h1>
Human cooperation is distinctly powerful. We collaborate with others to accomplish together what none of us could do on our own, share the benefits fairly, and trust others to do the same. These abilities are unparalleled in other animal species and are lacking in our most sophisticated artificial intelligence. What cognitive processes underlie our social intelligence, and how do these distinct abilities contribute to the scale and scope of human cooperation? How can this knowledge be used to design cooperative artificial intelligence that is equally capable of interacting with people? I will present a computational framework based on integrating individually rational, hierarchical Bayesian models of learning together with socially rational, game-theoretic models of cooperation. In computational models and multiplayer experiments, I will show how this framework can explain how cooperative behaviour is generalized: inferring the intentions and reputations of others, distinguishing friend and foe, and coordinating with AI agents and people.
<a href="https://faculty.washington.edu/maxkw/">Prof. Max Kleiman-Weiner</a> is is an Assistant Professor at the University of Washington in the Foster School of Business and Paul G. Allen School of Computer Science. He was previously a fellow at Harvard in the Data Science Institute and Center for Research on Computation and Society (CRCS). He completed a PhD in Computational Cognitive Science at MIT, advised by Josh Tenenbaum, where he was an NSF and Hertz Foundation Fellow. His thesis won the Robert J. Glushko Prize for Outstanding Doctoral Dissertation in Cognitive Science. He has received best paper awards at COGSCI and RLDM for models of human cooperation and the William James Award at SPP for computational work on moral learning. Max Kleiman-Weiner is a Co-Founder of Common Sense Machines and was previously Co-Founder and Chief Scientist of Diffeo, a collaborative machine intelligence startup acquired by Salesforce in 2019. Before all that, he was a Fulbright Fellow in Beijing, earned an MSc in Statistics as a Marshall Scholar at Oxford, and did his undergraduate work at Stanford as a Goldwater Scholar.
Date: Thursday, 20 February 
Location: Zoom (<a href="https://zoom.us/j/93978770269?pwd=ly42jZ5KnUiBHvz1pOeC9P3pUTYQLS.1">access here</a>) 
Time: 6:00 PM UK / 10 AM PT

alva-1-2

Alva 1

Come join us for this exciting event with CHIA ECC!
We have invited 2 speakers who will each give a talk about their research followed by a panel chaired by Alva Markelius. The speakers of this event are Justas Brazaukas and Minja Axelsson.
Justas is a PhD student at the Department of Computer Science and Technology and his primary interests span Sensor Networks-enabled Edge AI, Digital Twins, and Interaction Design. The topic of his doctoral project is Human-centric Design of Real-time Digital Twins with the thesis that digital twin deployments can be substantially improved by addressing the needs of different user segments, especially regarding the utilisation of sensors, the creation of data visualisations, and responsiveness to time-critical events.
Minja has recently defended her PhD thesis on robotic mental well-being coaches at the Department of Computer Science and Technology, University of Cambridge. Her current work is focused on the ethics of human-robot interactions, particularly in the context of well-being. Minja has previously been a visiting student at MIT Media Lab, and worked on robotic applications in industry.
Date: Tuesday, 4 February 
Venue: Syndics room, 17 Mill Lane, followed by drinks at The Anchor 
Time: 5:00 PM

whatsapp-image-2024-10-16-at-10-43-18-1

WhatsApp Image 2024-10-16 at 10.43.18 (1)

Flash talks, engaging discussions and nibbles!

Kicking Off the Early Career Community Winter Season
We’re excited to open our winter season of events with an evening filled with fascinating flash talks, lively discussions, and delightful nibbles! This is a great opportunity to share ideas, connect with others, and unwind in a friendly atmosphere.
We look forward to seeing you there!
Date: Thursday, 23 January 
Venue: The Castle Inn 
Time: 6:00 PM

img_2778

Giulia Occhini Seminar

Towards a Holistic Understanding of AI Language Readiness

Join us for the las ECC Autumn Seminar!  This seminar, by Giulia Occhini, will take place on 14 November at 6pm, in the Syndics Room, 17 Mill Lane. The seminar will be followed by informal drinks at The Anchor, offering a chance to continue the conversation.
In her talk, Giulia Occhini will explore the complex factors influencing the uneven deployment of NLP technologies. She will present new data and methods to better understand and address this issue, emphasizing the need for inclusive NLP development that promotes global equity. The talk will challenge traditional performance metrics in AI and call for a more responsible approach to ensuring that language models serve diverse global populations fairly.
Abstract: 
Despite its broad application in fields like healthcare and education, the benefits of NLP remain unevenly distributed. Indeed, most language technologies serve only a small portion of the global population, predominantly in the Global North. This disparity stems from the multilingual nature of the world, where high-quality NLP resources exist for only a small percentage of the world’s more than 7000 languages, exacerbating what is known as the digital language divide. This gap substantially limits access to essential knowledge and services for sustainable development, particularly for those communities and regions where such advancements are most needed. To adequately address inequalities in language technologies, we argue that we must first understand the complex factors influencing NLP deployment and assess the readiness of different languages and populations to engage with these technologies at the global scale. In this talk, we will propose new data and methods to disentangle the causes behind the unequal distribution of language technologies, stressing the need for inclusive NLP development to foster global equity. Achieving fair, global use of language models remains an open research challenge with significant ethical and practical implications, requiring a shift beyond traditional performance metrics toward more equitable, responsible AI.

b557c514-7625-41c5-96ed-dcbb8467c1e2_3456x1728

Joycelyn Longdon

Join us for an exciting Seminar on Ecoacoustics and Conservation!
The field of ecoacoustics has experienced remarkable growth in the last decade, driven by advances in machine learning and more affordable hardware. This progress has enabled the easier collection and analysis of large quantities of acoustic recordings, which are essential for biodiversity monitoring, research, and conservation. However, alongside these technological advancements, a critical issue has emerged—an imbalance in resources, expertise, and engagement between the Global North and South, particularly affecting Indigenous and Local Communities (IPLCs).
On Thursday, 7th November at 5pm in the Syndics Room, 17 Mill Lane, Joycelyn Longdon will delve into this complex issue in her seminar on ecoacoustics and its challenges in conservation. While ecoacoustics relies heavily on citizen scientists—particularly birding enthusiasts—for data collection and labeling, the communities most closely connected to the ecosystems being monitored are often left out of the conversation.
This seminar promises to spark important conversations about the future of conservation and how we can foster more inclusive, collaborative, and sustainable efforts to protect our planet’s biodiversity.
Abstract: 
Within the last decade, the field of ecoacoustics has seen accelerated growth due to advances in machine learning and reduced hardware costs that have facilitated easier collection and analysis of vast quantities of acoustic environmental recordings for essential biodiversity monitoring, research and conservation. Research in these fields has an established tradition of participation, supported by the expertise and labour of citizen scientists, primarily birding enthusiasts, in data collection or labelling tasks. Despite there existing significant overlap between regions of high biodiversity and the lands and livelihoods of Indigenous and Local communities (IPLC’s), like many other approaches to conservation, the field of ecoacoustics suffers from asymmetrical distribution of resources, expertise, equipment and engagement between the Global North and Global South, specifically, with local forest communities living in the ecosystems being monitored. This lack of engagement with forest communities and their beliefs and practices originates from colonialism and has been continued through neocolonial approaches, often resulting in tension between conservation practitioners, projects, and forest communities. This exclusion extends to the deployment and design of new ecoacoustic technologies, with Indigenous or Local Communities &#8220;seldom&#8230;involved in the technology design of conservation tools which results in the design of non-intuitive systems&#8221;, often leading to mistrust in the technology itself and its purpose for being introduced or deployed.  The rise in computational approaches to conservation and deployment of conservation technologies in forest landscapes threatens to compound and widen the disconnect between conservation projects and forest-dwelling communities, weakening important links to essential local knowledge and limiting the opportunity for effective, equitable and sustainable conservation interventions.

whatsapp-image-2024-10-16-at-10-43-14

ECC Seminar

Large and more instructable language models become less reliable

Our Early Career Community Seminars are about to begin!
We’re thrilled to announce that José Hernández-Orallo (Universitat Politècnica de València, and Leverhulme Centre for the Future of Intelligence), will be the speaker at our first seminar on Wednesday, 9 October at the Syndics Room, 17 Mill Lane.
José will introduce his recent study published in Nature. He will dive into the fascinating world of large language models (LLMs) and how their rapid scaling and evolution have led to incredible advancements. But here’s the twist—while LLMs are getting bigger and more sophisticated, reliability isn’t always improving.
Abstract:
The discrepancy between what is easy and difficult for humans and large language models (LLMs) has been well known since the first LLMs solved complex problems yet failed at very simple ones. This discordance with human difficulty expectations strongly affects the reliability of these models, as users cannot identify a safe operating condition where the model is expected to be correct. In 2022, Ilya Sutskever suggested that &#8220;perhaps over time that discrepancy will diminish&#8221;. With the extensive use of scaling up and shaping up (such as RLHF) in newer generations of LLMs, we question whether this is the case. In a recent Nature paper, we examined several LLM families and showed that instances that are easy for humans are usually easy for the models. However, scaled-up, shaped-up models do not secure areas of low difficulty in which either the model does not err or human supervision can spot the errors. We also found that early models often avoid user questions, whereas scaled-up, shaped-up models tend to give apparently sensible yet wrong answers much more often, including errors on difficult questions that human supervisors frequently overlook. Finally, we disentangled whether this behaviour arises from scaling up or shaping up, and discovered new scaling laws showing that larger models become more incorrect and especially more ultracrepidarian, operating beyond their competence. These findings highlight the need for a fundamental shift in the design and development of general-purpose artificial intelligence, particularly in high-stakes areas where a predictable distribution of errors is paramount.
Full article <a href="https://www.nature.com/articles/s41586-024-07930-y">here</a>.

anil-madhavapeddy

Planetary computing to help balance opportunity costs of human actions on wildlife

We live in a crowded world, where balancing the growing demands of food, fuel, and fiber for human populations against the urgent need to protect wildlife is increasingly challenging. How can technology help us strike this delicate balance?
Join us on Thursday 17th October, 6pm, at the S2 Room, Alison Richard Building, 7 West Road, as Anil Madhavapeddy (University of Cambridge Computer Laboratory) shares cutting-edge research on planetary computing and its potential to transform conservation efforts. Using satellite sensing and computational models, Anil will explore how we can measure biodiversity, track the impact of human activities, and even assess how global food choices influence natural ecosystems.
Abstract: 
The world is a really crowded place these days; we are trying to balance the food, fuel and fibre needs of a growing human population with the need to spare land for natural wildlife that are facing an extinction crisis. In this talk, Anil will discuss our ongoing work on building comprehensive maps of the world that measure important aspects of nature by leverage satellite sensing and computation pipelines. These models can measure things like tropical forest carbon, but also be composed to build the first globally comparable measure of biodiversity (through a species persistence approach), and also help measure the country-level impacts of human food choices to natural biodiversity. He&#8217;ll discuss some of the implications of this global monitoring approach: is this an opportunity to equalise the data landscape to the global south, or will it just reinforce existing data inequality?

illustration-of-a-smart-watch-with-health-tracking-features-being-used-by-an-elderly-person-the-watch-has-a-large

Illustration of a smart watch with health tracking features being used by an elderly person The watch has a large

The role of AI in cognition/altering human decisions or actions

The escalating mental health crisis presents a stark challenge to individuals and communities as well as healthcare and social services. The current system is overwhelmed leading to inadequate support, particularly in deprived areas, exacerbating health inequalities. With limited resources to hand, can machine learning, artificial intelligence, and robotics deliver the kind of care and treatment that we want and need? What safeguards should we put in place? What are the opportunities for innovation and enterprise?
Join the team from Cambridge University&#8217;s Centre for Human Inspired AI (CHIA) bringing together experts in psychiatry, cognitive neuroscience, robotics and computing for thoughtful discussion on one of the greatest health challenges facing society today.
&nbsp;
Schedule
09.15  Welcome coffee and registration
09.45  Conference opening by Prof. Trevor Robbins (University of Cambridge)
10.00  Keynote: Personalising antidepressant treatment in routine care: the PETRUSHKA tool. Prof. Andrea Cipriani, University of Oxford
10.30  Keynote: Artificial Intelligence and the Future of Psychotherapy. Dr Ana Catarino, ieso Group
11.00  Keynote: Social Robots for Assessing Child Mental Wellbeing. Prof. Hatice Gunes, University of Cambridge 
11.30  Coffee break
11.45  Panel discussion: Ethical and societal implications of AI for patients and mental health care. Chair: Catherine Galloway. Panellists: Prof. Andrea Cipriani, Prof. Hatice Gunes, Louise Croft (OGMA) and Valentin Tablan (ieso Group).
The panel discussion will be followed by a networking lunch reception.

eliseracine-thebigger-picturewebof-influence-i-2560x1746

EliseRacine-TheBigger PictureWebof Influence I-2560x1746

<h1>Collective Generative Futures</h1>
Join our first Early Career Conference 2025 to share insights and methodologies that push the boundaries of what is possible when humans and AI work together, aiming to shape a future where generative technologies serve as extensions of human capabilities rather than replacements.
In this half-day event, we will consider critical questions:
How can we put the synergistic potential of human-in-the-loop AI at the heart of our data-driven innovation?
How can AI and humans collaborate to enhance creativity, productivity and effective decision-making?
What ethical considerations must we address to ensure AI augments human life responsibly?
Together, we will examine the future of AI agents and robotics, discussing their integration into daily life, their autonomy, and the societal implications of such advancements. Through a keynote, panel discussion with colleagues from both academia and the industry, as well as presentations from the community members, we aim to envision a future where technology and humanity co-evolve to foster sustainable, inclusive, and dynamic societies. Join us to shape the narrative of our collective future with AI.
You will have the opportunity to present your work at the conference through either a 3-minute thesis or abstract presentation, with prizes awarded for the best contributions. Keep an eye out for our upcoming call for abstracts &#8211; your research and ideas will help shape this important discussion.
Please, complete <a href="https://docs.google.com/forms/d/e/1FAIpQLSefRbu2zv93Bw920j6uHBaKAWSBmi4CaOTKxwbkeeSwRAidBw/viewform">this form</a> to register your interest for attending our in-person conference. If you are interested in presenting, it will take you to the next page. If you have any questions, please contact the conference organisers:
Cherry Benson at <a href="mailto:cb2279@cam.ac.uk">cb2279@cam.ac.uk</a>
Abdullah Hasan Safir at <a href="mailto:sa2168@cam.ac.uk">sa2168@cam.ac.uk</a>
<h2>Schedule</h2>
<ul>
<li>1:00 PM – Arrival and Registration</li>
<li>1:15 PM – Conference start</li>
<li>1:25 PM – Keynote: Irmak Dogan</li>
<li>2:15 PM – 3-Minute Thesis Presentations</li>
<li>2:45 PM – Break</li>
<li>3:15 PM – Abstract Presentation</li>
<li>4:00 PM – Panel Discussion: Industry and Postdocs</li>
<li>5:00 PM – Pub Social</li>
</ul>
Date: 14 March 2025
Time: 1:00 PM – 5:00 PM
Location: Keynes and Chetwynd Rooms, King&#8217;s College, King&#8217;s Parade, Cambridge, CB2 1ST
Last Date of Registration: 5 March 2025
Abstract Submission Deadline: 28 February 2025
Picture: Elise Racine &amp; The Bigger Picture / Better Images of AI / Web of Influence I / CC-BY 4.0

14 March: Early Career Conference 2025

Past Events

18 February: Robotics and Embodied Intelligence Workshop

20 February: ECC Online Lecture

4 February: ECC Human centric AI design evening

23 January: ECC Welcome back event

ECC Seminar: Giulia Occhini

ECC Seminar: Joycelyn Longdon

ECC Seminar: José Hernández-Orallo

ECC Seminar: Anil Madhavapeddy

Conference: AI and Mental Health