Uses of Time

» The real potential is to use AI to take the pressure off teams, spending the time freed up to do the job better, take better care of relationships with colleagues, customers, suppliers, bring better ideas to the organisation, etc.

Karina Gibert is a leading figure in the field of artificial intelligence (AI) and data science. She is currently a professor at the Polytechnic University of Catalonia (UPC) and director of the research centre Intelligent Data Science and Artificial Intelligence (IDEAI-UPC), which she co-founded. Since April 2023, she has also been Dean of Catalonia’s Official College of Computer Engineering.

With a consolidated academic career, she holds a degree and PhD in Computer Science from the UPC, specialising in computational statistics and AI. Her research focuses on data mining, data science, explainable artificial intelligence, and intelligent decision support systems, with relevant applications in the fields of health, environment, public administration, and education.

She is also a prominent advisor to the Catalan and Spanish governments, the European Commission and the Commonwealth on ethics and AI issues. She is an active advocate of gender equality in STEM disciplines, and has promoted several initiatives to promote the presence of women in the technology sector, such as donesCOEINF and donesIAcat.

Throughout her career, she has received multiple awards, such as the Ada Byron Award 2022, the WomenTech Award 2023, and the National Computer Engineering Award 2023.

Her work focuses on ensuring that the development of AI is done in an ethical and responsible way, with special attention to transparency, sustainability, and social justice in the use of technologies.

1 – AI has become a central topic of technological and social debate. But how would you define it in an understandable way?

There are many definitions of artificial intelligence, but to understand the essence, I like to go back to its origins. The formal birth of AI occurred in 1956 at the Darmouth Summer School, a seminar organised by John McCarthy that brought together 10 experts (unfortunately, all men). McCarthy posed the following conjecture: ‘Is it possible to specify every aspect of human activity with sufficient precision that a machine can simulate it? And so AI (and these origins give the discipline its name) is set up as a mechanism for building machines capable of emulating human intelligence.

For a good handful of years, artificial intelligence focused on the development of automatic reasoning systems and expert systems designed to solve complex ‘thinking’ problems. This approach, however, went into crisis in the late 1980s, because it required a knowledge acquisition phase in which experts (doctors, mechanics, etc.) had to explain to engineers their knowledge of how they deduced a disease from its symptoms, or how they understood what was wrong with a car to repair it. It was found that humans do not know how to explain our thinking very well, since we make intensive use of implicit knowledge (acquired through experience and applied unconsciously), which is difficult to verbalise.

This limitation ushered in the AI winter, a phase of stagnation that lasted until the emergence of machine learning in the early 1990s. The new paradigm was revolutionary: instead of asking experts to describe how they reasoned, they were asked for data with real cases (symptoms and diagnoses, breakdowns and solutions, etc.) to build machines capable of learning as humans do.

Thus, it was not until the 1980s that data became the backbone of AI, initially as an instrumental tool to overcome the challenge associated with the implicit knowledge inherent in humans. Since the 1980s, we have witnessed an explosive combination of sensor and telecommunications technologies, which capture data with increasing accuracy, and AI methods, capable of learning increasingly complex patterns. This has given rise, among others, to filtering systems such as those of the Internet or social networks, which select the content they display according to the user’s profile or digital footprint, or predictive models that assess whether we will be good mortgage, health, car, or home insurance holders.

2 – And why is it now on everyone’s lips? We often make use of AI without being aware of it: content recommendations, health applications, virtual assistants… How long has it been living with us without us knowing it?

Since the 1980s we have been surrounded by ‘AIs’ that evaluate and assess us, some with more or less success, and some very beneficial, such as AIs that ‘see’ tumours in CT scans when the human eye still sees nothing. This somewhat erratic growth, however, has taken place against a backdrop of widespread deregulation.

The ‘generative AI boom’ consolidates in November 2022 with the free and open release of ChatGPT by OpenAI. A year earlier, DALL-E (textual description-based image generator) had already demonstrated the ability of generative AI to interact with non-technical users using natural language. The key difference was that DALL-E focuses on image creation, while ChatGPT is a conversational agent that responds to any request in human language.

ChatGPT’s intuitive and gamified interface, designed to create links, led to mass adoption because it breaks the barrier of the digital divide. It gathered 1 million users in one week and now has more than 800 million. This success catapulted generative AI into the media and the collective imagination, to the point that many users confuse ChatGPT with AI in general, even though it is only one of many generative AIs out there ,and generative AI represents only one AI. Not all AI is generative, and not all AI applications are made with generative AI.

3 – You have actively participated in the Artificial Intelligence Strategy of Catalonia, which promotes the incorporation of AI in key sectors such as health and welfare, public administration, the environment, agri-food, culture and mobility. From your experience, what concrete measures is this strategy starting to translate into, and what examples would you highlight that have a direct impact on citizens?

The Catalan artificial intelligence strategy (Catalonia.IA) was configured as one of the pioneers in Europe. The launch, planned for the end of 2019, was delayed due to the elections and was finally presented in February 2020, just before the pandemic. The strategy is structured through 4 instruments aimed at developing research (AIRA), innovation (CIDAI), adoption (DCA-IA) and all AI is developed from an impeccable ethical perspective (OEIAC), ensuring that the person and their rights are put at the centre.

CIDAI is perhaps the most active, with multiple lines of action. Among the different activities of the CIDAI, high-impact projects with applications for public administration are very valuable, including, for example, using AI to understand the impact of antibiotics on the quality and quantity of meat from fattening animals, or being able to better regulate traffic or better redistribute the bicycles of Barcelona’s Bicing service.

It also offers, for example, technical advice and proofs of concept subsidised by the strategy for SMEs and start-ups, guiding the introduction of AI into their businesses to increase competitiveness with minimised risks, because there is no initial cost for the entrepreneur.

4 – A recent CIS barometer shows a certain social scepticism towards AI. A significant part of the population is concerned about its possible negative consequences, especially in the labour market (Eldiario.es, March 2024). What would you say to the public to help them better understand the risks and opportunities of AI?

The catastrophic messages of the media do not help much, frankly. It would be necessary to modulate the discourse and how information is published. I am convinced that the evolution of the labour market does not depend on artificial intelligence or any emerging technology, but on strategic decisions in business policies and human resources management.

Workforces are undersized in all sectors. Taking advantage of the time that AI frees up workers to lay off is not the best decision, as it keeps job stress on those who remain. This choice does not lie with the technology, but with the companies. The real potential is to use AI to lighten the pressure on teams. This way, they can spend the time freed up to do the job better, take better care of relationships with colleagues, customers, suppliers, bring better ideas to the organisation, etc.

It is essential to have a minimal understanding of the foundations of artificial intelligence and to develop a critical view of how we consume AI daily, especially in a context where, for example, we receive hundreds of AI-based applications every day on our cell phones.

As for the technology gap, I think its days are numbered because with generative AI, the window opens to make AIs that anyone can talk to more or less naturally. With this interface, those who don’t make the technological leap will surely not be excluded.

I believe that the labour market, like all those that have faced disruptive technologies (from tractors to steam engines), will undergo an inevitable transformation. However, I am not currently concerned about the risk that large groups of people will be marginalised because they do not adapt. What worries me is how company policies will be oriented, but this is not a technological problem.

5 – It has been said that training is key. Do you think companies and institutions should actively invest in training their staff in AI, to take advantage of the potential and avoid a new digital divide? In this sense, what professional profiles would have to be trained beyond the technical ones?

Training is a key element. Within the framework of the Catalan Artificial Intelligence Strategy, we created the free course «ciutadania.cat», available online, designed to offer general knowledge about AI to any interested person. Currently, the course is being updated, and its material serves as a basis for training adapted to companies, which we often carry out.

6 – Another gap of concern is the gender gap. From your commitment to equality in the STEM field and as a promoter of initiatives such as donesIAcat or donesCOEINF, what is the current situation regarding women and AI?

I am deeply concerned about this. The under-representation of women in the technology sector is alarming. In Catalonia, if we exclude traditionally feminised roles such as customer service, marketing or human resources, only 9% of ICT professionals are women. In other words, if an organisation does not have at least 10 technology specialists, the technology it produces does not have a female perspective. And this is dramatic, because it means that we are building the 5.0 society (which is based on technology) without women and, therefore, their needs, particularities and even rights are not on the table like those of men when this future is being designed.

I find this so alarming that I work to network among women technologists to bridge this gap and inspire young women who want to train in this sector and dedicate themselves professionally. I have a real passion and devotion for this profession. Every day I ask myself how it is possible that there are not thousands of girls trying to be part of it. I work on projects of all kinds, collaborating with doctors, environmentalists, security forces, administrations, industries, etc. I am constantly learning about the issues that they face: water management, diseases, traffic, waste, etc. Understanding the challenges faced by each sector, I help to improve them by designing technological solutions, many with AI, although not always, and I see how the world is transformed when the solutions we have developed are implemented. It is very rewarding and interesting.

Now, moreover, those of us who know AI have become fashionable. Everyone is looking for us and needs us, but there are not enough of us to meet all the challenges and opportunities. We should be twice, three times… even ten times more. In Catalonia alone, 15,000 job vacancies will remain unfilled in 2023. Good, well-paid and fascinating jobs… How we don’t all want to be there is a mystery to me.

7- What dangers do you see if this technology is developed without a gender and diversity perspective?

The price of not including women in the tech sector is unaffordable, with many examples demonstrating the gender biases of AI. Women entrepreneurs have been systematically rejected by banks’ credit algorithms, trained with exclusively male historical data, which do not reflect the particularities of female entrepreneurship. Videoconferencing tools cut us badly and erase our eyes and ears because the models have mostly been fed with photos of young, white men. There are so many cases to explain… The solution is clear: equal development teams that integrate women in all phases of AI design.

8 – In the framework of the Barcelona Time Pact, we work to promote more efficient, equitable and sustainable organisational models that place quality of life, well-being and gender equality at the centre. From your point of view, how can AI help organisations to move towards this new time culture? Can you imagine concrete uses in the management of working time, flexible working hours or work-life balance?

The most prized value of artificial intelligence is its widely applied ability to absorb heavy and repetitive tasks that currently fall on people.

This should make it possible to lighten the workload, so that everyone can raise the quality standards of their work and no longer spend time on operational details. AI should be positioned as a «companion» or, a concept I like, as an «intelligent assistant».

Think of how women used to spend hours washing clothes by hand, with cold water and ashes, and today they are ‘assisted’ by washing machines that take on the heaviest work and gain time to do other things.

We should work towards a labour framework where AI plays the role of decongesting labour activity. This opens up many possibilities for professionals to devote their time to more human activities, such as a doctor who can spend more time with the patient, calmly, without the pressure of filling out digital documents, or a production line operator who can reduce the pressure because he knows that a machine detects faults better than he does and he only has to supervise the alerts issued by the machine.

9 – Looking to the future: what will organisations be like in 10 to 15 years’ time? What role will AI play in new ways of working and in the roles of employees? Experts such as Bill Gates suggest that we could move to 2-3 day working weeks. Is it realistic to imagine such a horizon, and how do we prepare for it?

I do not have a crystal ball to see the future, but I believe that the challenge ahead of us is immense. We are living through a change of era, of social and economic order, and the changes that are coming are profound and structural. It is essential not to confront them in any old way, but to reflect deeply on their consequences. All of us together have to make an exercise of depth and responsibility.

A tool like ChatGPT, which consumes vast amounts of energy and water and compromises the future of the planet, cannot be made available for everyone to play with or entertain themselves. Nor can we implement artificial intelligence overnight in an organisation without adequate preparation. Nor can we launch tools on the market that violate fundamental rights without oversight. These risks are not inherent to AI itself, but to the people who create, manage and use it.

If we do a good implementation of AI – and I find that we are quite aware of it here, especially compared to the United States or other parts of the world – we can protect the individual, improve the economy and advance ethically all at the same time.