Visions on AI at The Next Web Conference 2019

Author: Elmira van den Broek

What are visions of speakers and participants of the largest tech conference in the Netherlands on Artificial Intelligence (AI) and work? The KIN Center for Digital Innovation joined The Next Web Conference 2019 – two full days of workshops, panels, presentations and discussions around digital technologies – to find out everything around AI and the future of work.

The unforeseen consequences of AI

We started off with James Bridle (Author of The New Dark Age), who argued that technology is making the world harder to understand. According to Bridle: “the idea that we can understand the world by the gathering of data is starting to fail. The worst part is that we’re failing to notice. We see that in things that we tried to predict in the first place.”Bridle illustrated the bizarre and unforeseen events of AI with several examples, such as YouTube’s algorithmic bias towards misleading and harmful video content and Amazon’s algorithmically determined warehouse systems that replaced human intuition entirely.  Bridle suggested that to tackle these unforeseen consequences of AI, we should educate people in technology subjects, cooperate in better ways with technological systems, and distribute power and knowledge in technological networks. Perhaps, only then, will we be able to escape the dark age of the algorithm.

Educating children with VR and AI

André Kuipers (Astronaut at European Space Agency) showed how AI can be used for the greater good: the education of children. During his talk, Kuipers presented an innovative approach to bolster the realization of the limited resources of our planet. Based on personal footage and stunning visuals, Kuipers and his team developed an innovative Virtual Reality (VR) education program based on the latest VR and AI technology. The program aims to realize a permanent mind shift among children worldwide, by “impacting by experience”.

The role of AI in expert decision-making

During our roundtable session on AI and expert decision-making, we had the chance to ask tech professionals about their visions on AI in the workplace. During the session, roundtable participants argued that AI should always have a complementary role in decision-making: “The day we give up decision-making to AI, then we are on the wrong path”. Participants did not share the belief that AI would eliminate bias in the human decision-making process, but would only make bias less transparent: “If you delegate all the decision-making to these systems, you will not get rid of ungrounded biases, you will just not know what they are anymore”.They came to several solutions to facilitate fruitful collaboration between AI and experts. First, participants highlighted the importance of explanatory algorithms, as well as technological skills of experts to supervise AI-systems. Moreover, the use case of AI should be clear to the organization, and experts should be actively involved in developing AI-systems. “If it is us, the techies, that are going to be the ones deciding over the tech, which now is the case, then you would get the techie-perspective on a society built into your systems. And I don’t know if that’s a good idea”, a participant joked. The roundtable participants concluded that to make AI-systems work, professionals have the obligation to step in and participate in the development and use of these systems, in order to make sure they are used in the right way.

AI and the Workplace

We joined Hanneke Faber (President Europe of Unilever), who talked about purposeful brands and preparing the business for the future. In her discussion on how AI is being used at Unilever, Faber gave an example of the use of AI in recruitment: “We were pretty traditional in recruiting. You would apply online, and then you would get an e-mail back. [..] Today that is completely different. We have three AI-based rounds, before one last round where you would actually see a human”. According to Faber, the use of AI saves both the applicant and recruiters an enormous amount of time: while it used to take applicants 3-4 months to complete the selection process, now applicants know in 2 weeks whether they are hired or not. “But the really cool thing for us is that it has significantly increased the diversity of the people we hire”, Faber emphasized. Therefore, Unilever perceived the use of AI as an important means to increase diversity in the workplace.

Combining AI and Human Intelligence

According to Rand Hindi (Co-Founder and CEO of Snips) we do not have to fear a future with AI. In his talk, Hindi highlighted the difference between “narrow AI” – doing one task, without the ability to generalize – and “general AI” – doing any logical task, using reasoning and intuition. Hindi explained that it will be difficult to move from narrow to general AI because emotional human intelligence cannot be easily replicated into machines. This is the case because humans are considered conscious beings, human emotions are core to decision-making, knowledge begins with our senses, and intelligence is collective.  According to Hindi, we can, therefore, expect that the future will all be about “the combination of the emotional power of the human brain and the logical power of the artificial brain”.In sum, the combination of AI and human intelligence will be our best bet.

Ethics and Biases in AI

Finally, we joined Lydia Kostopoulos (Senior researcher at Digital Society Institute), Rob McCargow (Director of AI at PwC), Jane Zavalishina (Co-Founder of Mechanica AI), and Lofred Madzou (AI and Machine Learning Project Lead at World Economic Forum) for a panel discussion on ethical AI. According to the panel participants, problems with AI and ethics arise because ethics is contextual, and we are aiming to create AI-systems that are more ethical than human beings. Participants identified different solutions to enhance ethics in AI, such as devising ethical metrics or KPIs, having diversity in AI-development teams, leaving room for “unknown unknowns”, and reflecting on ethical values and the use case of AI. These solutions should help organizational leaders to take the path towards ethical AI.

In sum, The Next Web Conference taught us some of the promises and drawbacks of the use of AI, and that among others collaboration, education, transparency, and reflection on aims and values are crucial for realizing beneficial and ethical AI-systems in practice.