Welcome to this month's SSI Fellows Newsletter which shares activities and opportunities taking place within the SSI Fellows' community. Read on for:

• Fellows' Spotlight: Daniele Procida, Gemma Turon and Hannah Williams.
• November 2023 Community Call recap
• Actions for Fellows
• Fellows' and related activities
• Upcoming events and calls

Fellows’ Spotlight

Daniele Procida, Director of Engineering, Canonical

Updates from call

• This year I took my work on documentation https://diataxis.fr to data scientists in southern Africa (February) at PyCon Namibia. I also had a deeper research community encounter at FAIRmat in Berlin, where a colleague and I ran a two-day documentation workshop for the team behind NOMAD (this engagement continues). 
• At other events, I spent time with SREs and others working on research software to discover their challenges. The theme that has emerged from the year is: the product model of software transfers well to research software, and seems to clarify its documentation problems effectively.
• Not only that, but the response I had to “Practices for quality at scale” workshops I’ve run this year suggest that pragmatic and lightweight methods to advance good engineering practice in general at the organisational level also seem to have value for research contexts.

My non-work highlight - I was part of the organisation of the first-ever DjangoCon Africa, which took place in Zanzibar in November. We started planning the conference in 2018, only to be derailed by successive global and regional events. The outcome was a huge success: 200 people gathered for six days, from 22 countries, and another step forward for open-source grassroots organisation in Africa.

My recommendations - I finally read the “new” (i.e. 2011) Bill Johnston translation of Stanisław Lem’s Solaris. It’s still only available in digital editions, and far more lucid than the much earlier Polish to French to English translation. It must be the funniest novel about the history of science I’ve ever read. It’s also painfully sad (about human beings) and on the idea of the possibility of being able to understand non-human intelligence, still more deeply insightful than anything I’ve seen in the last year in the froth of conversation about LLMs.

Gemma Turon, Co-founder and CEO, Ersilia Open Source Initiative

Updates from call - The SSI Fellows call is a fantastic opportunity to reflect all the work done in the context of the Fellowship (whether it has been economically supported by the Fellowship or not). In the latest call, I presented briefly what our young non-profit organisation, Ersilia, is trying to achieve in the field of data science for infectious disease research. With the goal of reducing inequalities in global health, we are building a set of open-source Artificial Intelligence tools to support scientists working in low resourced settings. Thanks to a huge community effort, our main platform, the Ersilia Model Hub, now features over 120 AI models for drug discovery, and covers disease areas like malaria, tuberculosis, schistosomiasis and HIV. 

My non-work highlight - I’ve started working on my solo circus act. As my side-job and long-time passion, I perform and teach aerial acrobatics. I have started working on my first long solo show, with the idea of premiering at the end of 2024!

My recommendations - The latest Cannes Palme d’Or awarded movie: Anatomy of a Fall

Gemma’s slides from the Fellows Community Call can be found here.

Hannah Williams, Research Software Engineer, Dstl (previously, and whilst completing fellowship activities, Mathematical Modeller, UKHSA)

Updates from call - During my fellowship period (2021) I worked with colleagues to establish a UKHSA-wide Software Community of Practice. This was a larger undertaking than my original proposal, but (hopefully) a more comprehensive forum to facilitate sharing and learning best practice and internal knowledge. We hosted seminars, coffee-and-coding, workshops and built a network for one-on-one support.

• Unexpected time-consuming activity: talking to people; ensuring that teams/individuals would support the community, and securing senior approval.
• Could have done better: measuring impact. 
• Biggest mishap: collaborative documents going wrong.
• Highlight: the opportunity to learn from people (huge shout-out to the SSI Community Building study group)

My non-work highlight - A holiday (8 years in the pipeline) with a friend to Disney World Florida. It was even more than supercalifragilisticexpialidocious. The highlight of the trip was Fantasmic; a musical-pyrotechnic-laser-water show.

My recommendations - My go-to podcast for my commute is “It’s a Fair Cop” by Alfie Moore (available on BBC Sounds) – highly entertaining and informative insight into life as a police officer, giving the audience an opportunity to decide what course of action they would take if they were in Alfie’s shoes.

Hannah’s slides from the Fellows Community Call can be found here. 

Community Call Recap

During the November 2023 SSI Fellows Community Call, we heard Fellows’ updates from Daniele Procida, Gemma Turon and Hannah Williams. In breakout rooms, we discussed dealing with management, RSE & Python (especially in LMIC), and handing over things.

Fellows’ and related activities

Collaborations Workshop 2024 - The Collaborations Workshop 2024 (CW24) will bring together researchers, developers, innovators, managers, funders, publishers, policy makers, leaders and educators to explore best practices and the future of research software. The themes of CW24 will be: Environmental Sustainability, AI/ML tools for science, Citizen science

• Date: 30 April - 2 May 2024
• Location: Scarman, University of Warwick
• Link: https://www.software.ac.uk/workshop/collaborations-workshop-2024-cw24 

Upcoming events and calls

• RSE Conference 2024 - The eighth annual conference for Research Software Engineering will take place in Newcastle. See the link below for more details and submission deadlines.
  • Date: 3-4 September 2024
  • Location: Newcastle, England
  • Link: https://rsecon24.society-rse.org/ 

Everyone has read at least one paper lately that started “Machine learning has made an impact in the entire field”.

Doesn’t matter which field.

Then ChatGPT happened, and even those peddling crypto and web3 suddenly talked about AI.

We've got AI hype on top of our AI hype.

But with that, a few conversations became disproportionately harder.

The Emergence of Alignment

It happened when chatGPT was released.

People, including myself, couldn’t believe how good this type of model had become. The texts were fluent. Depending on your instructions, there was what seemed like coherent thought.

It seemed a bit like magic, or maybe even gifted with consciousness to some.

Machine learning papers had already been published without peer review at a break-neck pace before chatGPT was released. Some models, like the popular Yolo models, were even released completely without formal publication.

So, publishing claims that chatGPT showed sparks of Artificial General Intelligence, without peer review and as quickly as possible, fit into the culture of machine learning research.

It’s a compelling narrative.

What if we created an intelligence? Wouldn’t that at least be cool?

But quickly, questions emerged about the long-term safety of humanity. If we created superhuman intelligence, how do we ensure it doesn’t turn against us? How do we ensure AI is aligned with humanity's goals or even survival?

The idea of AI alignment was born.

The sudden interest in Long-term Ethics

Suddenly companies with big well-paid AI teams, especially those that recently fired their entire ethics teams, were talking about the long-term view.

AI alignment was everywhere.

These companies were even pushing for regulation of these AIs and future development of AIs. Some researchers started comparing themselves to Oppenheimer. Quite the emotive comparison. Let’s ensure the long-term survival of humanity by regulating the alignment of AI development.

Yet behind the scenes, those same companies pushed for de-regulation behind closed doors in the EU, where the AI Act would have wide implications.

It’s compelling, though. We all want to survive. We’re all a bit scared of the long-term effects of this un-known entity. It’s the Siren call. Or it’s a good set-up for a trick.

The Magic Trick

From street magic to David Copperfield, magicians all work with one simple trick.

Attention.

A magician will direct your attention to the hand they don’t want you to look at. Then they can perform the sleight of hand with the other, non-observed hand.

When we get enough people to pay attention to the long-term harms and be satisfied with the regulation of long-term goals, we can divert the attention from short-term impacts.

Not too dissimilar to other industries that have spent resources to divert our attention from lung cancer and climate change.

What are we missing in the Short Term?

In the short term, we know machine learning can already do harm.

From surveillance to increased premiums. Rejected loan applications that can’t be explained by random blips in The Algorithm. Biases in machine learning models and algorithmic decision processes are well-documented.

And then there’s, of course, the CO2 footprint of AI model training, which will be an actual proven problem in the future. We don’t have to speculate in this case.

These biases and discriminatory correlations can be present in the training data, exacerbated through exploratory data analysis and then reinforced by humans through automation bias. And then there is, of course, the loss of privacy through widespread data collection practices.

Something that already affects us every day and could be addressed by data cleaning practices and regulations. (Coincidentally, why Google Bard is not available in the EU right now.)

Scientists using Machine Learning

As researchers, this goes even deeper.

The use of machine learning in different fields of science has been proven extremely potent. However, many basic tutorials don’t adequately reflect the realities of real-world data.

It can be computer vision models that recognise the hospital code of places that treat patients that were advanced cases, therefore cheating through this shortcut. Or it could even be researchers training on the test data to improve the score, beat the state-of-the-art and be able to publish. There are accidents and questionable behaviours that can be hidden in machine learning scores.

These publications can chart the entire trajectory of a field, like we unfortunately recently discovered with fabricated data in Alzheimer’s research.

We have a responsibility for AI safety, even if it’s just our tiny little field.

Why I created ML.Recipes

I wrote the resources ML.recipes to empower scientists to quickly learn about best practices.

The Jupyter book has code examples everyone can easily re-use and explanations throughout the entire book as to how to work with this information and the benefits of each section.

These are generally divided into “ease review” to make it easy for others to gauge the quality of a machine learning contribution, “increase citations” to be able to verify and re-use your code and model, and “foster collaboration” which is about standards and share-ability of your science.

Let the billionaires and effective altruists discuss AI alignment while we build useful tools and insights to advance science and humanity. Tools that are safe and science that is valid.

Large language models may not be our end just yet.

After an hour of discussion on a research software related topic at a workshop, a discussion group would stand up for 2-3 minutes and present back their findings (e.g. problems, solutions, future work or however they chose to speak about a topic). However without context the notes produced from such a session are not of much use to the wider community after the workshop. So, what's the solution?

Fueled by adrenaline and a shared passion for innovation, seventy-eight university students from across Latin America converged virtually for the second edition of the Co-Afina hackathon. Armed with open data and computational tools, these bright minds tackled real-world challenges related to the environment, citizenship, and education, showcasing their skills, motivation and collaborative spirit.

This year's Co-Afina hackathon was made possible by the support of the International Center for Theoretical Physics (ICTP) and its Physics Without Frontiers (PWF) programme, the Software Sustainability Institute (SSI) and Open Life Science (OLS), as well as the artificial intelligence company INAIT SA. Joining forces with the organising alliance of the LA-CoNGA physics project, Creative Commons Venezuela, the Latin American Advanced Networks Cooperation (RedClara), and the Academic Network of Ecuador (CEDIA), these sponsors provided the platform for Latin America's next generation of innovators to shine.

Over three intense days (October 20-22, 2023), sixteen teams, including students from thirty universities across eight Latin American countries, immersed themselves in the hackathon's challenges. With creativity and determination, they devised solutions that addressed pressing issues such as informed voting, market forecasting, and sustainable tourism.

The winning solution, "Electoral Radiography of Colombia," emerged from the minds of the "DarkArmy" team, representing the Universidad Industrial de Santander (Colombia). This innovative web application aims to empower citizens with informed voting decisions by providing a comprehensive overview of electoral candidates' profiles. The team members, Jorge Jaimes, Camilo Carvajal, Brayan Barajas, and Paula Uzcátegui, showcased their exceptional problem-solving skills and commitment to civic engagement.

Second place went to the "Chaotic Coders" team from Yachay Tech University (Ecuador). Their proposal, "Predicting the Value of Coffee," employed machine learning techniques to forecast the future value of coffee in the financial market, leveraging historical data to make informed investment decisions. The team members, Arianna Paredes, Mateo Carpio, Jordan Zambrano, and Kevin Robalino, demonstrated their proficiency in data analysis and financial modelling.

Rounding out the top three was the "CongApps" team, a multinational alliance of students from the Universidad Industrial de Santander (Colombia), the Universidad Simón Bolívar (Venezuela), and the Universidad Nacional Mayor de San Marcos (Peru). Their solution, also focused on promoting informed voting, presented an accessible app for citizens to gain insights into electoral candidates' profiles. The team members, Steven Rico, Adrian Vasquez, and Omar Medina, highlighted the importance of transparency and accountability in the democratic process.

Beyond the top three, the "Neotropical" team received a special mention for their proposal, "Quantum Peru." This environmentally conscious project sought to mitigate the impact of tourism on archaeological and culturally significant sites, demonstrating a deep understanding of sustainable practices. The team members, Andrés Caña, Cristian Usca, Emilio Toledo, Rubén Niño, and Isabella Sánchez, represented universities from Venezuela and Ecuador.

Participants and organisers expressed excitement over the innovative solutions developed during the hackathon. Yany León Castañeda, a Co-Afina challenger, highlighted the results achieved in a short time, emphasising the enthusiasm, knowledge, and creativity of the participants. The organisers commended the teams for their critical, creative, and environmentally sensitive perspectives.

Co-Afina 2023 showcased the incredible work of Latin American student teams, emphasising their ability to respond swiftly and effectively to regional challenges. The organisers expressed gratitude for the collaborative effort that made this edition possible and concluded the event with an invitation to next year's hackathon. With this call to action, Co-Afina sets the stage for continued innovation and collaboration among Latin America's brightest minds, paving the way for a brighter future.

Let me start out a bit vulnerable. 

I have never been particularly popular, and as a millennial, that led to me seeking out online communities.

It turns out that was partly rooted in my neurodivergence, and so it carried through my life. So, when the SSI pushed for more communities of practice around our fellowships, I felt like I should give it another try.

Ironically, in the current AI hype, I'm still inclined to think that I don’t talk about things people would find interesting or would like to talk about with me. Maybe something for the therapist’s office.

But how do communities form?

Even for me, making friends in school was easy. The person you got sat next to ended up being your friend in a lot of cases.

So young adults often reach a cliff-like drop in connection when they move to their first job that leaves them isolated and curious: what happened? Why is this so hard?

So instead of proximity, modern communities form around shared interests.

Online, we see this a lot around three incredibly popular topics: money, relationships, and fame.

The only problem? You have to deal with people that only care about money, relationships, and fame.

So instead, we have to put on the hard work and go from “Zero to Not Zero” when we start a football club, just as much as when we build a new RSE community.

What is the hard work?

Creating a community is vulnerable.

As someone that opens the community, you bare a part of your soul to the world and hope it is received well.

But certain aspects can break your community. When we start with a lack of diversity and grow the community, we will forever play catch up, for example.

Other problems can be personality cults, where someone makes the entire community about themselves. And as humans, we are very clearly flawed, and that flaw is very often the downfall of a place you spent years in.

Then there are poisoned wells, where we are scared to exclude people through moderation, and suddenly a person or group can spread their toxicity in this place you used to call your home.

And then there’s the whole part about getting people into the door, which is its own crux.

What works?

Is there any saving?

Well, many places have successfully built diverse and inclusive communities. They have a few things in common:

Codes of Conduct work as a promise to at least try and care about the safety of a community. They lay the foundation to have something enforceable by community leaders. But that obviously lives and dies with how the leadership in a community adheres to and enforces this code.

Strong moderation to avoid the poisoning of your well, but easy access so folks can join without too much of a hassle. Different studies on usability show that each question on a form drastically reduces the probability that a user will sign up.

At the Collaborations Workshop, we had the wonderful “Pacman Rule”. Be welcoming and always leave a spot open if someone would like to join a circle of conversation. Being welcoming usually sets the tone for a community.

There’s also value in having some neat merch like stickers, but that can also be a bit much to organise, especially in the beginning.

Don't go alone!

You'll start alone.

But get keen people on board as quickly as possible. The old saying goes, “Alone, you go fast. Together, you go far.” That way, your community isn’t hinged on you doing well alone but actually spreads the work to multiple people.

Someone said to me that building a community is training others in leadership to take your community further.

What I learned from discussing and building

Choose what you care about and pursue it with integrity.

Include your kind of people.

Go with software that people actually use. I tried to use a fancy “community software” to build The Latent Space. This ended up being a bit of a waste because people struggled to sign up. Emails weren’t sent. And the interface was clunky. Now we use Discord, a software that is battle-tested and fairly easy to set up and definitely easy to moderate.

Another consideration for inclusivity as a community founder is proper certification. You can’t become a Certified Best Friend™ but you can definitely get safety certifications like the Mental Health First Aider certification to deal with certain situations adequately.

What is the Latent Space?

If you’re curious, The Latent Space is an inclusive machine learning and AI community.

I myself am queer and neurodivergent. I try to embody feminist and black-inclusive values, but those aren't my identity, so I educate myself and implement them as best as I can.

It is for practitioners and the curious. Whether you want to talk about machine learning, recommender systems, data science, AI, chatGPT, or the ethics of all of it, this is the space.

Whether you like to enthusiastically share things you found or engage in deep discussions about the impact of new developments, this should be a place for both. Just a place for "the normal stuff". The things below the hype. But the rest will have to grow organically.

Remember, building communities is hard, so getting started is brave!

SSI Fellow Professor Tom Crick MBE has assumed the position of Chief Scientific Adviser (CSA) for the Department for Culture, Media, and Sport (DCMS). In his new capacity, Professor Crick will offer scientific and technical guidance within the department, deliver direct counsel to ministers and officials, and supervise the application of research, evidence, and external expertise. Collaborating closely with other CSAs in various government departments across Whitehall, under the leadership of the Government Chief Scientific Adviser, Professor Crick will ensure the implementation of a unified and strategic approach to tackle key cross-cutting government policies.

Professor Tom Crick expressed his delight in joining the DCMS and eagerness to collaborate with colleagues to advance the department's commitment to being data-driven and analytically rigorous. He is committed to integrating robust science and engineering methods into both policymaking and the overall departmental culture, as well as contributing to innovative, place-based approaches that maximize the potential of their sectors, nurture creative communities, and create diverse opportunities for young people. Additionally, he mentioned his desire to comprehend the long-term impact of artificial intelligence on culture, heritage, and the creative industries. He is also eager to work with Chief Scientific Advisers across government on broader strategic science and innovation priorities, fostering connections with the wider research community in academia and industry, both within the UK and internationally.

We are delighted to announce that SSI Fellow Professor Tom Crick MBE has been awarded the Lovelace Education Medal by BCS, The Chartered Institute for IT for his contributions to computer science education across research, policy and practice.

Tom Crick is Professor of Digital Education & Policy and Deputy Pro-Vice-Chancellor at Swansea University. He is recognised internationally for leading the major STEM education and skills reforms in Wales over a sustained period, alongside wider leadership in UK digital, engineering and technology policy to support a thriving digital and data-driven economy.

Alongside Tom Crick, Demis Hassabis CBE, co-founder of AI research company Google DeepMind, and computer scientist Professor Jane Hillston MBE have been announced as the receipents of the Lovelace Research Medal.

The BCS Lovelace Medal was established in 1998 in honour of mathematician and writer Lady Augusta Ada Byron, Countess of Lovelace. The award recognises people whose work in the areas of research and education have contributed to significant advances in computing. Previous winners include worldwide web inventor Sir Tim Berners-Lee and information retrieval pioneer Karen Spärck Jones.

SSI Fellow Michael Tso is among the organisers of a community workshop focused on the environmental code of long-term value. This free hybrid event, which is tailored for both NERC data centres and environmental modellers, is scheduled for Tuesday 21 November and will be conducted online as well as in person at UKCEH, Lancaster.

The main goal of NERC data centres is to ensure that environmental data are made available, accessible, and reusable for the long term to fully realise their value. Environmental data includes data collected in the field, samples analyzed in the laboratory, as well as model outputs that are underpinned by model code. Model code also plays a vital role in data analysis, understanding and visualisation, and decision-making. Thus, it is essential to consider model code in the same light as datasets regarding their long-term value. However, safeguarding the long-term value of model codes presents unique challenges that are not found in datasets. Model codes are prone to frequent changes, they may become unusable when software dependencies become obsolete, and ensuring their quality can be difficult.

UKCEH and the EIDC, which is part of NERC Environmental Data Services (NERC-EDS), are actively engaging with the environmental research community to understand the user community's needs concerning model codes. They are seeking input to foster ideas that can help both users and developers of environmental codes maximize their long-term value. Your input can also contribute to shaping future NERC strategies, policies, NERC EDS commissioning, and the development of the NERC model metadata catalogue.

 The workshop will cover a range of important themes, including defining what constitutes an 'Environmental Model Code of Long-Term Value,' exploring best practices and standards, addressing training and upskilling needs, delving into emerging principles and ideas, such as research objects, and discussing how to capture the long-term value of model codes. This will include considering the levels of FAIRness and metadata requirements.

Furthermore, the event will examine the role of data centres and long-term repositories in preserving these vital resources and provide a roadmap for the way forward.

If you are interested in participating, please fill in the registration form by Tuesday 14 November. T&S support is available.