Question 1
How would you explain AI to a 5-year-old?
Asked to explain AI to a five-year-old, the contributors described machines that learn complex tasks from examples rather than fixed instructions, making mistakes and correcting them from feedback. Tirthajyoti Sarkar of Rhombus Power called modern AI a prediction machine that finds hidden patterns in data and, like Alex Wang of Deloitte, separated today's Narrow AI, trained for a single task such as recommending a film, from General AI, which did not yet exist. Siddha Ganju of NVIDIA used things a child already knows: a computer learning what a cat looks like, Snapchat placing dog ears on a face, and Alexa understanding a request.
“AI is supposed to be a human being’s ‘intelligent companion.’”
Tirthajyoti Sarkar, Sr. Director, Rhombus Power“Using Artificial Intelligence we can teach a computer to think and reason what makes a cat look like a cat.”
Siddha Ganju, Senior Artificial Intelligence Lead, NVIDIA, NASA Frontier Development Lab, SpaceML
Answered in the book by Mike Tamir, Tirthajyoti Sarkar, George Panou, Alex Wang and Siddha Ganju.
Question 2
What is the essential curriculum for a career in AI?
Most contributors set out the same base: linear algebra, probability, statistics and programming, then hands-on projects and the domain knowledge school rarely teaches. Michael Wu of PROS called AI inherently interdisciplinary, its best practitioners working with lawyers and designers, and said non-technical people such as ethicists and doctors were needed too. Several mapped the roles, from data engineers to ML engineers and researchers, with a PhD needed mainly for research. Cynthia Rudin of Duke University stressed interpretable machine learning and ethics, both rarely taught, and Jon Skirnir Agustsson of Nox Medical said those who stood out paired a strong foundation in one field with skill at working across disciplines.
“AI is inherently an interdisciplinary subject.”
Michael Wu, Chief AI Strategist, PROS Inc.“AI is just a tool and for it to have a real impact it should be applied to the right challenges, at the right time, and in the right manner.”
Jon Skirnir Agustsson, VP Artificial Intelligence and Data Research, Nox Medical
Answered in the book by Michael Wu, Aruna Kolluru, Mike Tamir, Cynthia Rudin, Tirthajyoti Sarkar, Anand Ranganathan, Jon Skirnir Agustsson and Siddha Ganju.
Question 3
What is the most impressive task AI can accomplish today?
Aruna Kolluru of Dell Technologies chose computer vision, whose accuracy had risen from 50% to 99% since 2012 and which was helping diagnose medical images, while others cited language and speech, recommendations and drug discovery. Jordi Guitart of Aizon said the most impressive task was still to come, in the diagnosis and prognosis of life-threatening conditions, and warned that many companies had paid heavily to tackle poorly defined problem statements. Cynthia Rudin of Duke University cited models that predicted server failures, grid fires and seizures, but warned that AI was also being weaponised, through misinformation and facial recognition built on scraped personal photos.
“The most impressive task that AI can accomplish is yet to come and it will deal with Human health.”
Jordi Guitart, VP of Artificial Intelligence, Aizon“AI is helping society in numerous ways, though as we know, it is also being weaponized.”
Cynthia Rudin, Professor of Computer Science, Duke University
Answered in the book by Cynthia Rudin, Jordi Guitart, Jyotirmay Gadewadikar, Aruna Kolluru and George Panou.
Question 4
How is AI impacting the way businesses operate?
Simon Greenman of Best Practice AI called AI a general-purpose technology already used across sectors and functions, best suited to frequent, repetitive, data-rich tasks with clear inputs and outputs, and said it needed board oversight and a governance framework. Edward Pyzer-Knapp of IBM Research saw AI moving businesses from data-driven to discovery-driven, citing faster materials discovery, and Raj Madan of Arcutis Biotherapeutics pointed to diagnostics and drug discovery in healthcare. Several wanted people to keep the final say, with the means to monitor and override AI, and stressed explainability, bias detection and data privacy.
“Ultimately AI is a technology that requires Board of Director oversight as it can lead to brand reputational damage along with significant financial damage if it goes wrong.”
Simon Greenman, Partner & Member of WEF Global AI Council, Best Practice AI“AI is key in the evolution of business from a data-driven to a discovery-driven enterprise.”
Edward Pyzer-Knapp, WW Research Lead, AI Enriched Modelling and Simulation, IBM Research
Answered in the book by Edward Pyzer-Knapp, Andy Pardoe, Simon Greenman, Raj Madan and John Thompson.
Question 5
How could a company transform into an AI-powered organisation?
Richard Benjamins of Telefónica set out a sequence others echoed: explore a few use cases, then break down silos, appoint a Chief Data Officer and build a data team over three to five years, and only then use AI at scale. Most advised starting with use cases that matter to the core business yet are feasible, proving value to senior leaders and scaling step by step, and Christian Guttmann of the Nordic Artificial Intelligence Institute wanted the first projects to win over sceptical stakeholders. Jyotirmay Gadewadikar of MITRE called it naive to expect results from AI tools alone. Views split on buying AI services or building in-house teams for differentiation.
“It is naive to think that enterprises’ efficiency and efficacy can be improved just by adopting AI tools.”
Jyotirmay Gadewadikar, Chief Engineer, Enterprise AI, MITRE“Initial AI projects or products should aim to be meaningful, rather than impactful, for as many stakeholders as possible.”
Christian Guttmann, Executive Director, Senior Researcher, Nordic Artificial Intelligence Institute, Karolinska Institute
Answered in the book by Jyotirmay Gadewadikar, Richard Benjamins, Harry Mamangakis, Yasen G. Dimitrov, Aruna Pattam, Jordi Guitart, Sunil Kumar Vuppala, Christian Guttmann and David Pereira.
Question 6
Do companies need a Chief AI Officer?
Most contributors wanted a senior executive accountable for data and AI but split on the title and reporting line. Richard Benjamins of Telefónica held that the Chief Data Officer should own both, since AI depends on data, while Jordi Guitart of Science4Tech Solutions argued the two roles were not interchangeable and that a Chief AI Officer was essential wherever AI is the main source of competitive advantage. John Thompson of Marketing Sciences called placing analytics inside IT the original sin and wanted it reporting to the CEO, COO or strategy. Others saw the role evolving with maturity, from a Data and AI team under the Chief Data Officer.
“The best corporate role for being accountable for data and Artificial Intelligence is the Chief Data Officer.”
Richard Benjamins, Chief AI & Data Strategist, Telefónica“The common mistake organizations make is that they put the analytics team in the IT organization.”
John Thompson, Author, Principal Consultant, CEO, Marketing Sciences, LLC
Answered in the book by Andy Pardoe, Harry Mamangakis, Agnis Stibe, Richard Benjamins, Jair Ribeiro, Jordi Guitart, Sunil Kumar Vuppala and John Thompson.
Question 7
What is the AI state of the art?
Dima Turchyn of Microsoft described transformer-based models as having reshaped natural language processing, with very large models such as the 530-billion-parameter Megatron-Turing NLG improving results across many tasks, and new models learning from text and images together. He noted that the time from a new model to its use in products was shrinking, which raised the need for responsible use. Mike Tamir of Susquehanna International Group said large language models did not necessarily understand text yet had moved far beyond rules, and pointed to gains in reinforcement learning. Emma Duckworth of GSK Consumer Healthcare described AI orchestrating global supply chains and noted that transfer learning had commoditised computer vision.
“The cycle from new model introduction to implementation significantly shortens.”
Dima Turchyn, Artificial Intelligence Product Marketing Lead, Microsoft“I am consistently impressed with how far we have come in natural language understanding and processing more generally.”
Mike Tamir, Chief ML Scientist, Head of ML/AI, Susquehanna International Group
Answered in the book by Eva Agapaki, Dima Turchyn, Emma Duckworth, Netanel Eliav and Mike Tamir.
Question 8
What is AI's impact on our societies and everyday life?
Contributors expected large gains, from removing repetitive work and raising productivity to earlier diagnosis and help with epidemics, pollution and food shortages, and many set these against bias, misinformation, loss of privacy and power concentrated in a few technology firms. Michael Wu of PROS warned of large-scale job displacement in the short term and of growing inequality. Nell Watson of the IEEE Standards Association expected large language models to deskill much of software development. Alf Rehn of the University of Southern Denmark described a ‘long bias’: because AI favours what is known, it may cut speculative ideas, so people must keep the capacity to imagine.
“Many aspects of software development will be significantly deskilled, or perhaps bifurcated.”
Nell Watson, Tech Ethicist, Researcher, Reformer, IEEE Standards Association“It is of critical importance that we retain the human capacity to imagine and dream.”
Alf Rehn, Professor of innovation, design, and management, University of Southern Denmark
Answered in the book by Michael Wu, Nell Watson, Anthony Mills, Alf Rehn, Marily Nika, Nazar Zaki, Angeliki Dedopoulou, Jon Skirnir Agustsson and Boyka Simeonova.
Question 9
What is the impact of AI on employment?
Alexandre Guilbault of TELUS argued that past technological revolutions transformed roles without causing mass unemployment, and that AI would first reach sectors with readily available training data, leaving manual trades such as plumbing among the hardest to change. Harry Mamangakis of Voiceweb saw AI as a decision-making tool that would replace some unskilled, repetitive work, citing a McKinsey estimate that under 5% of jobs would be lost but around 60% would need to adapt. Mike Tamir of Susquehanna International Group expected AI to speed up the removal of repetitive work, from serving coffee to long-haul trucking, but thought it could make knowledge work more effective rather than devalue it.
“AI helps professionals work more efficiently and make better decisions.”
Harry Mamangakis, Chief Technology & Operations Officer, Voiceweb“We should all educate ourselves so that we can benefit from AI to improve and facilitate our work.”
Alexandre Guilbault, Director of Advanced Analytics & AI, TELUS
Answered in the book by Alexandre Guilbault, Harry Mamangakis and Mike Tamir.
Question 10
How should societies get prepared for AI?
Golestan Radwan, who led Egypt's national AI strategy until 2022, called AI education possibly the most important aspect of adoption, tailored to each society: awareness programmes that dispel myths, compulsory courses for domain experts such as healthcare and agriculture professionals, and training for leaders in treating AI as an investment. Mathew Hughes of Loughborough University warned that AI could learn bias from its data and, since society is too dispersed to oversee it, looked to government, business and non-government institutions to make it serve the public good. AI ethicist Himanshi Allahabadi wanted education systems to teach how AI works, its reliance on historical data, and its limits and risks.
“A vital component of a society's readiness for the age of AI is to train its ‘domain experts’.”
Golestan Radwan, AI Researcher and Strategist“Society must not become a passive observer of Artificial Intelligence.”
Mathew Hughes, Schulze Distinguished Professor of Entrepreneurship, Loughborough University
Answered in the book by Golestan Radwan, Mathew Hughes and Himanshi Allahabadi.
Question 11
What are the ethical concerns of AI?
Most of the eleven contributors put bias learned from historical data and human choices at the centre, and several said it could be reduced through representative data, diverse teams and subgroup testing, but never fully removed. Veena Calambur of Workday warned that AI could automate biased decisions at scale under the guise of objectivity. Therése Svensson of IBM placed the true risk in uninformed or careless use rather than the singularity, and proposed that every model carry a content declaration of its intended use and limitations. Enrico Panai of Éthiciens du Numérique argued that the fear lay in the unpredictability of consequences, not the speed of AI.
“Without intervention, AI algorithms can learn, codify and perpetuate biases long into the future.”
Veena Calambur, Data Scientist & AI Ethicist, Workday“To ensure proper use, AI systems should come with a content declaration describing the intended use and limitations.”
Therése Svensson, Data Science & AI Ethics Solution Specialist, IBM
Answered in the book by Arathi Sethumadhavan, João Azevedo Abreu, Aruna Kolluru, Veena Calambur, Enrico Panai, Mike Tamir, Aruna Pattam, Therése Svensson, Samiran Ghosh, Netanel Eliav and Mayte Hidalgo.
Question 12
How do we prevent the concentration of AI power?
Abhishek Gupta of the Montreal AI Ethics Institute traced the concentration to compute, data and talent: training large language models cost millions, leaving the research and product agenda to a few well-funded labs. He supported a government-funded national research cloud, well-maintained open datasets such as Mozilla's Common Voice, and community training programmes. Netanel Eliav of Sightbit saw data as the fuel of AI and looked to regulation to stop large platforms collecting data that people did not share, since governments exist to correct market failures. Both saw meaning and purpose, beyond pay, as the way for smaller players to draw talent away from Big Tech.
“To draw AI talent, provide meaning and purpose beyond just compensation.”
Abhishek Gupta, Founder and Principal Researcher, Montreal AI Ethics Institute“It is extremely difficult for small companies to attract AI talent”
Netanel Eliav, Chief Executive Officer, Sightbit Ltd
Answered in the book by Abhishek Gupta and Netanel Eliav.
Question 13
To regulate or not? How should governments react to AI?
Most contributors favoured regulation, but Jaroslav Bláha of trivago argued that AI needed no more or different regulation than classical software, only better requirements, testing and quality assurance. Simon Greenman of Best Practice AI, once a sceptic, saw regulation as the answer to AI's trust problem and pointed to the risk-based approach of the proposed EU AI Act and to mandatory explainability statements. Kashyap Kompella of AI Profs noted that product safety, consumer protection and anti-discrimination laws already applied to AI, and that the bar should rise with the cost of errors. Others proposed a global treaty, a consequentialist approach to lawmaking or ethics by design.
“High-risk AI use cases should not stop technology from being deployed.”
Simon Greenman, Partner & Member of WEF Global AI Council, Best Practice AI“Trust in AI systems is a necessary condition for their adoption.”
Kashyap Kompella, CEO, AI Profs / RPA2AI Research
Answered in the book by Jaroslav Bláha, Yannis Kalfoglou, Luca Sambucci, Gertjan Boulet, Aruna Pattam, Badr Boussabat, Simon Greenman, Enrico Panai, Samiran Ghosh and Kashyap Kompella.
Question 14
How could democracy benefit from AI?
Joseph Yun of the University of Pittsburgh blamed engagement-driven recommendation algorithms for pushing users towards ever more extreme, personalised views of reality, and proposed models tuned for movement towards common ground instead. Mike Tamir of Susquehanna International Group said machine learning was unlikely to become a truth detector but could often tell when the author of a text had an agenda. Enrico Panai of Éthiciens du Numérique argued that democracy was too difficult to reduce to computable parts: AI could help with its complex processes, but not choose representatives or run the state. Others saw AI helping citizens build political literacy and trace the sources of fake news.
“Expecting an ML algorithm to become a magic truth detector might be naïve.”
Mike Tamir, Chief ML Scientist, Head of ML/AI, Susquehanna International Group“The reality of a democracy is so difficult that it cannot be reduced to a sum of elements.”
Enrico Panai, AI and Data Ethicist, Éthiciens du Numérique / Beethical.Be
Answered in the book by Joseph Yun, Enrico Panai, Agnis Stibe, Mike Tamir and Jennifer Victoria Scurrell.
Question 15
Can AI help solve climate change?
Both contributors saw AI as a strong tool for climate action that carried its own carbon cost. Prakash Natarajan of GSK Consumer Healthcare described it as a Swiss army knife for sustainability, from more accurate climate projections to the circular economy and grid integration of renewables, and urged organisations to cut their own AI emissions and share models rather than duplicate them. Eleni Mangina of University College Dublin stressed that technology alone was not the solution: using AI called for a cost-benefit approach, changes in energy behaviour, and orchestration by global organisations that address concerns about the data collected.
“AI can be described as the ‘Swiss army Knife’ for sustainability and climate change solutions given its multi-functional capabilities.”
Prakash Natarajan, Head of Sustainability, Tech Consumer Healthcare, GSK Consumer Healthcare“AI is a tool to help us fight climate change and we need to follow a cost-benefit approach, given that AI has its own growing carbon footprint.”
Eleni Mangina, VP (International) Science, IEEE Senior Member, University College Dublin
Answered in the book by Prakash Natarajan and Eleni Mangina.
Question 16
How is AI changing the global geopolitical system?
Contributors agreed that AI would shift the balance of power between states. Badr Boussabat of AI Together argued that power would go to states that build digital giants fed by the data of large, connected populations, while small states should specialise in niches such as semiconductors. Simon Greenman of Best Practice AI expected China to lead in applied AI because of its scale, and saw the world heading towards a split between Chinese and US influence. Luca Sambucci of SNGLR Group warned that AI could give authoritarian states what they had lacked, the means to make sense of the data they collected, and would transform the nature of warfare.
“AI will rebalance the power between states, upend traditional economic models, and transform the nature of warfare.”
Luca Sambucci, Head of Artificial Intelligence, SNGLR Group“Geopolitics will inevitably be structured around the development of AI.”
Badr Boussabat, President, AI Together
Answered in the book by Agnis Stibe, Badr Boussabat, Simon Greenman and Luca Sambucci.
Question 17
Will AI become capable of replicating or redefining itself?
Mike Tamir of Susquehanna International Group called general AI controversial but not inconceivable, and suggested that advances in machines' handling of language might be closely related to it. Andy Pardoe of AGI Wisdom Group expected AI, once general intelligence was reached, to design improved versions of itself, saw AutoML and generative design as early signs, and argued that explainable AI would keep humans from being left behind. Jaroslav Bláha of trivago noted that self-replication was no sign of intelligence and that self-improvement had so far stayed within a single domain; he judged general AI a distant goal.
“As with lowly biological systems, the ability to replicate is no indicator of intelligence or consciousness.”
Jaroslav Bláha, CIO/CTO, AI Evangelist, Trivago N.V.“The problem for us is that AI is likely to evolve much quicker than we will be able to unpick and understand its designs.”
Andy Pardoe, Founder & Group CEO, AGI Wisdom Group
Answered in the book by Jaroslav Bláha, Andy Pardoe, Mike Tamir, Steve Shwartz and Sherin Mathew.
