以下是Natasa Milic-Frayling 的演讲实录：   

As a platform for convening the leading talent and presenting cutting-edge achievements in the field of artificial intelligence, the Global Artificial Intelligence Technology Conference (GAITC) is a strategic meeting of international importance, focusing on the latest AI developments in the world. The conference is held annually and focuses on the theory, technology and application of artificial intelligence. It offers to attendees a feast of excellent academic and industry work that promotes innovation and integration of artificial intelligence into a wide range of disciplines. 

Natasa, Chair of Data Science at the University of Nottingham's School of Computer Science, is a prominent scholar in the field of data analytics and information systems, with 17 years of research experience at Microsoft Research in Cambridge, England. She holds PhD in Mathematics from Carnegie Mellon University in Pittsburgh. 

[At the start of her professional career in Computer Science, she conducted research in Information Retrieval at the Clairvoyance company in Pittsburgh.  Since then, she has diversified her research in all aspects of information systems and brings rigour and systematic approach to the work on algorithms, system design and user experience.] 

Natasa promotes a holistic approach to technology and innovation by adopting the ecology paradigm to set a research agenda and ensure a rigorous and systematic approach to the design and evaluation of systems, algorithms, and user experience.   

At the GAITC in Nanjing, Natasa spoke about “Artificial Intelligence in the Midst of the Modern Media Ecology and Social Transformation” . She discussed how information ecology can be applied to understand in depth the transformational impact of AI by considering four main pillars: people, practices, values   and technology. Natasa provided illustrations of specific issues are important to consider.

First, it is critical to identify main stakeholders in the media ecology and identify areas where the interests of news consumers and news producers may align or may differ.  Who are we designing for? What are we providing and how does that affect other stakeholders? These are the questions that computer professionals grapple with.  A recent study, conducted through collaboration of University of Nottingham and University of Zagreb, revealed that both the consumers and producers of news highly appreciate quality news. However, there is a tension in their preferences.  While media companies aim to provide personalization and their unique commentary on the reported events, most of the readers desire ‘objective truth’. In search for that ‘truth’ they consider reports from multiple media providers. They visit different news sites to read different reports and form their own opinion. That behaviour is in direct conflict with the aims of media companies to secure the loyalty and attention of their readers. 

Related to the same issue, editors and reporters of news may have different priorities when publishing news.  News editors focus on readership and cater to the interest of a specific community. They are likely to use sound bites as titles of stories.  Reporters, on the other hand, object to that practice on the ground that such titles can be misleading, particularly in the era when users may not have time to read the full story. 

Second, Natasa talked about social media engagements that both the media producers and the media consumers take part in. The sheer scale and the speed of communication make it difficult for individuals to get a sense of the online community, its global structure and their role in it. The question is whether we can provide tools that non-computer scientists can use to gain insights and make informed decision how to engage. Natasa illustrated the use of NodeXL, an extension of Excel that is distributed by the Social Media Research Foundation and aims at non-technical users analyse and visualize relationships among participants in their social networks. The NodeXL social maps of XinhuaNet followers showed the most recent activities on Twitter and the groups of followers that connect on specific topics that XinhuaNet community talks about.  

Third, Natasa raised an important issue of trust in technology and how it is undermined by the lack of transparency in the design of software applications. The issues is particularly critical when it concerns widely used applications such as the Web browsers. Browsers typically fail to expose real time tracking through cookies that the users are constantly exposed to. Visitors of news sites are likely to be tracked by third parties associated with the news companies through advertising. Natasa showed the intricate and elaborate network of trackers that involve a large number of sites and track users as they browse from one site to another. While real-time tracking may provide benefits to all involved, the end users have been completely disempowered, not given a choice to decide about their participation. It is not surprise that the consumers’ trust in the technology and practices has been affected. 

Finally, Natasa illustrated how AI methods can be used in marketing and sales that involve social media and e-commerce platforms.  Through ongoing collaboration with JD’s team in Beijing, Natasa’s PhD student, Weiqiang Lin has explored one of the key questions: How effective is a company’s communication with fans and followers on social media in inspiring interest in products and encouraging sales? Through analysis of social media content, it became clear that the social media conversations are not a consistent predictors of product sales. In fact, the performance of models trained to predict sales can be seen as a way to measure how aligned the social media content is with the consumer willingness to buy. 

Examples that Natasa described confirm that the media ecology is, indeed, an effective framework for reasoning about the transformational impact of AI on research and innovation. Each reveal the intricate interactions among the stakeholders and their individual role in shaping the technologies, practices and values that we observe. 

（本报告根据速记整理）