MIT Technology Review Arabia
AI is embedded in all aspects of life - financials, work, leisure, and social activities. Following this trend forward, how will AI artificial intelligence continue to impact daily life? How might its trajectory be better directed towards solving the issues and concerns of today, such as public health and the environment? Further, what is the role of data governance, and how will augmented intelligence facilitate the management of these challenges?
To answer these questions, Majarra and MIT Technology Review hosted a virtual conference entitled “Augmenting Life '' on November 30, 2021, with Microsoft as presenting partner and PwC knowledge partner for the conference. The conference was host to top international and regional experts, researchers, and entrepreneurs representing global organizations and educational institutions such as Microsoft, Google, Electronic Arts, PwC, Fetch.ai, Boston University, the Cary Institute of Ecosystem Studies, among others.
Opening the conference was Dia Haykal, Director of Brand and Partnerships at Majarra, who in her introductory keynote discussed the importance of predicting the impact of AI, its applications, and benefits on various sectors, highlighting the importance of facing these subjects early on in the Arab world through hosting global experts and thinkers.
The floor was then given toMario Vargas, EMEA Advanced Analytics' Lead at Microsoft, who presented a keynote entitled “Breaking down barriers to AI adoption”, in which he explained the main reasons holding organizations and companies back from adopting AI technologies. These included fear of the unknown, fears related to privacy and cybersecurity, and not enough awareness or understanding of such technologies.
Vargas then set forward a starting point in adopting AI, revealing a study stating that 91% of companies will have started using AI within the next three years. He also indicated one more obstacle, however, related to existing policies and mentalities currently in practice, which require the provision of the necessary skills development, sufficient financing, and clear vision for the adoption of such technologies. He also indicated that business leaders must themselves understand the importance of making sense of this technology before they can properly encourage their employees to adopt and use it responsibly, underlining the importance of establishing a culture ready to integrate AI into their companies.
Beginning the set topics for the day was Barbara Han, Disease Ecologist from the Cary Institute of Ecosystem Studies in New York, touching on the central role AI plays in reshaping the future of public health, and in particular the role that big data and machine learning can play when predicting pandemics.
Han offered attendees a glimpse of her research, presenting a previous study - for example - where machine learning was used to predict the type of vermin that may transmit disease to humans, filtering out 244 species from an original 2277 that can potentially transmit disease. She explained that there were a number of factors that would make such vermin carries, such as size, maturity, geographical distribution, and others. Another study was able to establish a machine learning model that could predict the animals that could transmit ebola with an accuracy rate of 80 - 90%.
When the Coronavirus pandemic spread, Han and her team developed AI models to predict areas most likely to spread the pandemic, studying the likelihood that the virus would attach to certain cells from various animal life, and estimate the environments where the virus could spread by attaching to the largest number of hosts. She concluded her keynote by clarifying that it is currently impossible to perfectly predict the time and place of a new virus of such potential, but what can be done is estimate risks and monitor the areas in which it is possible such a new pandemic could occur.
Sercan O. Arık, Senior Research Scientist and Manager at Google, was next during his keynote, “AI-Augmented Epidemiology for Covid-19”. He stated that precise prediction for the spread of the pandemic carried great benefit for major health organizations, policy makers, manufacturers, business leaders, and the public in general. To this end, Arik and his team had developed a platform based on AI to process these predictions, making them available in the US and Japan during 2020. Among the most important features of this platform, he clarified, is that it is dynamic, amending its predictions based on different changing variables such as precautionary measures taken and availability of intake spaces at hospitals, and interpreting its findings to policy makers.
Kicking off the discussion sessions for the day was a panel featuring Ahmer Inam, Chief Artificial Intelligence Officer (CAIO) and Chief Product Strategist at Pactera EDGE, and Tarek Khorshed, Lead Technology Architect in the WHO Geneva Bureau, moderated by Shaikha Al Othman, Founder/CEO of Haus of Care.
Khorshed, speaking in his personal capacity, stated that AI centers around building smart systems that can learn from offered data to help us make better decisions, seeing this ability as holding a special importance in the health sector, and stating an example from his research using machine learning to predict cancer diagnoses. He explained that it is usually difficult to diagnose this disease before it reaches advanced stages that are difficult to treat, and therefore using AI to allow for early prediction could potentially save patients’ lives. He considered the greatest challenge to adopting AI is the availability of health data to be used when training models.
For his part, Ahmer Inam - the first to coin the term “Mindful AI” - discussed the current contradiction between what AI can currently offer and its potential on the one hand, and the practical adoption of this technology on the other. He explained that human nature requires that leaders show empathy when designing appropriate solutions that can be commonly and sustainably adopted. He also stated that we must think of the reasons and motivations behind adopting AI whenever completing any project. For the health sector, he felt that this would mean developing effective management systems capable of providing the appropriate care at the right time and with the right price.
The topic’s second discussion panel, “Using AI Tools to Make Health Services More Accessible, Affordable, and Inclusive”, was moderated by Ahmad Nabeel, Founder & CEO, Gulf Medical Technologies, with Scott Nowson, Director of Digital Services at PwC Middle East, and Karim Dakki, CEO and Cofounder of KLAIM both speaking. Nowson stated that he is working on helping PwC clients define the positions in which AI can be put to best use in their companies. For the health sector, he clarified that his company is working with service providers, project managers, and others to provide consultations on how to make use of AI when offering auditory assistance, for example, or establishing the correct technology for computer visualization, machine learning, or psychological application development. Nowson considered AI “a solution searching for a problem”, indicating its great and promising potential.
Explaining KLAIM’s specialization in employing AI to serve the health insurance field, Dakki emphasized during the same discussion session the need to use the proper case studies to apply this technology and necessity for the appropriate data when training the machine learning models. He explained that the journey of AI application is a long one, and requires a lot of trial and error to verify the ability of the models. He shed light on the importance of protecting data privacy in the health sector, as well as maintaining the standards followed in this field.
Concluding the first topic of the day was a keynote by another representative of the knowledge partner, Matthew White, Partner - Digital Trust at PwC. In “Responsible AI: How important is it?”he stated that there were several dimensions that still require investigation in order to ensure the application of AI responsibly. Most importantly: ethics and regulation; bias and fairness; interpretability and explainability; robustness, security, safety, and privacy; and governance. He delved into some of the risks concerning privacy invasion, data theft, racial or ethnic discrimination, and increased injustice in the healthcare sector which must be handled when considering applying AI.
AI’s impact on the environment
The second session of the conference covered the views and research of experts regarding leveraging AI in facing climate change, in addition to conserving energy, and lowering the carbon footprint of training current machine learning models.
Leading the charge with her keynote: “Addressing the High Energy Costs of Learning in Artificial Intelligence Systems”, Kate Saenko, Associate Professor and Director of Masters in AI at Boston University, presented the methodology for machine learning, artificial neural networks, and the reasons that they consume massive amounts of energy, as well as offering possible solutions to this issue. Saenko clarified that the performance of these networks improves as they increase in size, and that training artificial intelligence models leads to carbon emissions equal to developing and using 5 cars throughout their entire life cycle. A single application of BERT, the natural language processor, for example, generates the same carbon emissions as the travel of one individual by air from San Francisco to New York.
With the costs clearly identified, the conference continued to propose specific solutions to the issue with: “Leveraging space intelligence to fight climate change”. Ed Mitchard, Co-founder and CTO of Space Intelligence, stated that there were tens of thousands of artificial satellites circling the earth, around 1000 of them gathering data on the surface of the earth. Most of this data is available to everyone and easy to access. Mitchard explained his view that these images can help in reforestation efforts, and discovering areas of deforestation that would be difficult to reveal from the surface of the earth. It would also be difficult for humans to process all this data, which is why Mitchard applies AI to process and extract practical data, such as knowing which areas have suffered deforestation.
How AI is transforming entertainment and culture
How AI will reshape the future of entertainment, culture and art was the topic at hand for the third and final session of the day.
Humayun Sheikh, CEO and founding partner of Fetch.ai, started the session by proposing a problem and its solution: how can we encourage people to help train cultural AI without permanently handing over priceless cultural artifacts and proprietary materials? To answer this question, he began by explaining that his company is working on developing the infrastructure needed to employ artificial intelligence in automating complex duties by relying on multi-client systems, decentralized machine learning, and secured data sharing.
Sheikh clarified that decentralized machine learning systems are being built using blockchain technology which allow the sharing of machine learning features among data proprietors. He stressed the importance of making these AI tools available in an open source format to maximize the benefit, using as an example tools applied in the transport and delivery field allowing a direct relationship between the service provider and the client without resorting to a mediator. Finally, Sheikh discussed the CoLearn platform which allows various parties to work together to train machine learning models without sharing data from each with other parties when participating in the training.
Delving further into the philosophical, Chris Ume, Co-Founder of Metaphysic.ai and the person behind the deep fakes of US actor Tom Cruise that were widely spread throughout social media previously this year, presented a session entitled: “The future of Synthetic media, its creative use cases”. In it, he clarified that the production of these videos took him several months of work in order to feed the model huge amounts of video and photography data for the actor Tom Cruise. Ume indicated the need to acquire the approval of any famous personality before the use of deep fake technology when generating an image that will be used in an advertisement, for example. He also covered another aspect of deep fake, which is faking the movement of lips and the speech generated, technologies which have been used when dubbing movies and advertisements, and even when producing new music using the voices of deceased artists. The capabilities of this technology in filmmaking, particularly when de-aging the faces of actors, as well as other applications regarding the generation of video game characters that approximate the true features of footballers - for example - in a FIFA game, requires creating awareness in society about the risks of misusing such technology.
Representing an industry reaching 3 billion players worldwide in 2021, of them 434 million gamers in the MENA region alone, Uma Jayaram, General Manager of SEED at Electronic Arts (EA), discussed the impact of gaming on a wide variety of people and the use of AI in current and future games during her keynote: “AI in Game industry: Building New Worlds and New Mindsets”.
Jayaram also clarified that gaming starts with a challenge, followed by a choice, then a change, and finally an opportunity. AI has wide applications in each of these stages, as it can be used in automating game flow, and in creating new features and experiences such as open world building, improving gameplay, and increasing the speed of launch. She also displayed some AI technologies in use currently for game development, such as enhanced learning, competitive regenerative networks, and artificial neural networks.
Finally, with the last keynote for the conference, Jason Bailey, Co-Founder and CEO of ClubNFT, discussed the future of art production and verification during his session entitled: “Can machine learning predict the price of art at auctions?” One example he offered was of a company scanning paintings using machine learning to expose fraud, clarifying that that is directly linked to estimating value of artworks. He also clarified that machine learning revealed the success of artists is greatly connected to how close they are born to global art centers, and continues to research the value of the painting as related to where the artist created it and the number for those willing to buy it, which currently has a much stronger connection to its success over the quality of the artwork itself.