Wetin the Artificial Intelligence (AI) Regulation mean for Silicon Valley
Julia Reinhardt, Privacy Management and Artificial Intelligence Professional, on data policy, the General Data Protection Regulation (GPA) and the impact wey e get on SMEs
In the last episode of Voices of the Data Economy, we follow Julia Reinhardt talk. He be expert in the management of Artificial Intelligence and a consultant in the field of privacy and public policy based in San Francisco. As a Mozilla Fellow in Residence, Julia dey assess the opportunities and limitations of European approaches to dey Reliable Artificial Intelligence in Silicon Valley and their potential for US business and defense. During our discussion, Julia talk about the different aspects of the impact of the GCC in Silicon Valley and the challenges of the forthcoming Artificial Intelligence regulation. Following na edited excerpts from the podcast.
(English and Greek Subtitles)
Impact of GCC in Silicon Valley
The GCC get significant and immediate impact worldwide. It don also make Silicon Valley aware sey privacy dey important to people in Europe and around the world. Na human right wey many in the United States no consider important. The global debate over privacy dey shifted over the last three years from 2018 onwards, as have laws. As a direct result of GCC in Europe, countries such as Japan, Brazil, India and China dey in the process of passing GPP-inspired privacy laws. In addition, California get new privacy law wey go into effect in 2020 due to the GCC.
GKP also show Silicon Valley wey one of its largest markets, Europe, get their own rules and sey the US must follow them in order to be a major player there and make money in the region. As a result, many US-based organizations wey dey process personal data of people around the world don decide to implement the GCP and extend all the rights wey accompany it to their clients, who no need to dey resident in Europe and live outside Europe. . This go give them an edge in global compliance and make am easier for them to handle complaints and requests. In addition, the GCC provide them with a legal framework and a set of standards.
"I go talk sey that one dey frustrating factor with GCC laws is their enforcement. Even when technology companies dey hit with billions of dollars in fines, for them na a blow to the wrist. And so far, the GCP nave change the subjects business models, how to make money online and monitor people's behavior, no be just a company's business model - na the whole internet business model - it no get have privacy To change this requires fundamental and possibly painful adjustments to the way things dey structured, something wey the GCC get so far don failed to achieve. And na definitely a bit of a disappointment."
Artificial Intelligence (AI) Regulations in Europe and their Global Challenges
Julia don work as a German diplomat for almost 15 years, manage bilateral relations, crisis management, communications, leading high-level protocols, dey participate in EU negotiation processes and dey promote innovation and scope in the western US.
In the context of its current work, e dey state sey e intend to ensure sey the forthcoming Artificial Intelligence (AI) regulation from Europe no go again cause this lag for young players, because for the field of Artificial Intelligence (AI) - the size matters. "The more data you fit gather, the better your Artificial Intelligence (AI) system go work. luxury to collect high quality data, wey then allow them to fit build Artificial Intelligence (AI) with better performance. And for small-scale providers, wetin dey most important na the clarity of guidance. The draft wey dey submitted by the European Commission don prepare for a long time. Na the most ambitious and comprehensive effort to reduce the risks associated with the development of Artificial Intelligence (AI) wey we don see so far around the world. This na bold new step. "You fit read an analysis of the proposed Artificial Intelligence (AI) regulation here.
Now, for 2021, we dey in the process of translating these principles into practical rules and regulations. The rules dey proposed by the European Commission no go cover all Artificial Intelligence (AI) systems. For example, them cover systems wey dey considered to pose a significant risk to the security and fundamental rights of people wey dey live in Europe. This na a risk-based approach and has many levels. And these levels dey different rules for different categories of Artificial Intelligence (AI) systems - some dey banned, some dey considered high risk and some only follow certain rules. And then others where, you know, them just say you need to be more transparent.
Deepening the rules: Code tests for algorithms
You need to know which category your Artificial Intelligence (AI) system falls into. For some uses of Artificial Intelligence (AI), the Commission dey propose a total ban and says it is an unacceptable threat to citizens. For example, an Artificial Intelligence (AI) system dey cause physical or psychological harm by to manipulate people's behavior or exploit their vulnerable abilities, such as age or disability. Other examples na social grading systems where people collect points and face recognition for public by law enforcement - not all face recognition dey prohibited, but those wey dey used by the police in public places. Although exceptions dey.
Most of the master plan dey focus on Artificial Intelligence (AI), wey dey considered high-risk, and that high-risk dey defined in the master plan. Thus, these na in some ways get problematic uses in the field of recruitment in the context of hiring, to determine a person's creditworthiness or eligibility for public services and benefits, as well as some applications wey dey used in law enforcement and security and judicial system. For them, too, these systems must meet different requirements and undergo conformity assessment before being placed on the European market.
To ensure sey an Artificial Intelligence (AI) system dey comply with various requirements for the management of serious risks, na must use data sets in training, validation and testing wey dey relevant, representative, error free, etc. The documentation for a high-risk Artificial Intelligence (AI) system go really extensive and very accurate - why you choose specific designs? Why you design am that way? The key word dey always human supervision. Thus, high-risk Artificial Intelligence (AI) systems must dey designed to enable people to understand the capabilities and limitations of a system and to fit address the so-called automation bias. Also, if necessary, to reverse or bypass the result. E be like to control code for algorithms.
Windows in the setting of Artificial Intelligence (AI): It is not the final world
The European Parliament and other bodies in Europe don already call for much stricter rules on some elements of the plan. In addition, some Member States believe sey e go be stricter in some cases. However, this no be the final reason.
"In my personal opinion, I think the exceptions, like in face recognition, dey very broad. E dey difficult when you ban specific uses of face recognition, but then, in industry or in private uses, prohibition no dey. Even In law enforcement uses, there some areas where e fit dey used dey. Practically, law enforcement authorities for Europe go purchase face recognition systems on the market, wherever them dey produced, and use them in specific cases where them dey allowed. You really wan make sure them no use them for other things? I think sey this na huge window in. I think sey face recognition get the potential to really undermine our free society." So lot to criticize go dey in this plan."
The following na list of selected timestamps for the various topics wey dey discussed during the podcast:
2:16 - 6:56 Julia's journey from a German diplomat to a consultant on data policy and regulations in the US.
6:56 - 12:48 The impact of the GCC in Silicon Valley
15:55 - 20:38 Impact of the GCC on the Technology Giants and Small and Medium Enterprises in the USA
20:38 - 25:53 How do the proposed Artificial Intelligence (AI) regulations affect the US?
25:53 - 29:58 Detailed analysis of Artificial Intelligence (AI) regulations
29:58 - 35:55 Window and Challenges of the Artificial Intelligence (AI) Regulation?
35:55 - Finally Innovation and the regulation of Artificial Intelligence (AI) go hand in hand?