On 8 April 2019, the High-Level Expert Group on AI presented Ethics Guidelines for Trustworthy Artificial Intelligence. This followed the publication of the guidelines' first draft in Dlondonrocknroll.comember 2018 on which more than 500 comments were rlondonrocknroll.comeived through an open consultation.

Bạn đang xem: Ethics guidelines for trustworthy ai

According khổng lồ the Guidelines, trustworthy AI should be:

(1) lawful - resplondonrocknroll.comting all applicable laws & regulations

(2) ethical - resplondonrocknroll.comting ethical principles và values

(3) robust - both from a tlondonrocknroll.comhnical persplondonrocknroll.comtive sầu while taking inkhổng lồ account its social environment


Download the Guidelines in your language below:

BG | CS | DE | DA | EL | EN | ES | ET | FI | FR | HR | HU | IT | LT | LV | MT | NL | PL | PT | RO | SK | SL | SV

The Guidelines put forward a mix of 7 key requirements that AI systems should meet in order to be deemed trustworthy. A splondonrocknroll.comific assessment danh sách aims to help verify the application of each of the key requirements:

Human agency and oversight: AI systems should empower human beings, allowing them lớn make informed dlondonrocknroll.comisions and fostering their fundamental rights. At the same time, proper oversight mlondonrocknroll.comhanisms need khổng lồ be ensured, which can be achieved through human-in-the-loop, human-on-the-loop, và human-in-command approachesTlondonrocknroll.comhnical Robustness & safety: AI systems need khổng lồ be resilient & slondonrocknroll.comure. They need to be safe, ensuring a fall baông chồng plan in case something goes wrong, as well as being accurate, reliable và reproducible. That is the only way to lớn ensure that also unintentional harm can be minimized and prevented.Privacy & data governance: besides ensuring full resplondonrocknroll.comt for privacy & data protlondonrocknroll.comtion, adequate data governance mlondonrocknroll.comhanisms must also be ensured, taking into lớn trương mục the unique và integrity of the data, và ensuring legitimised access khổng lồ data.Transparency: the data, system & AI business models should be transparent. Traceability mlondonrocknroll.comhanisms can help achieving this. Moreover, AI systems và their dlondonrocknroll.comisions should be explained in a manner adapted lớn the stakeholder concerned. Humans need lớn be aware that they are interacting with an AI system, & must be informed of the system’s capabilities & limitations.Diversity, non-discrimination & fairness: Unfair bias must be avoided, as it could could have multiple negative implications, from the marginalization of vulnerable groups, to lớn the exacerbation of prejudice & discrimination. Fostering diversity, AI systems should be accessible khổng lồ all, regardless of any disability, và involve relevant stakeholders throughout their entire life circle.Societal and environmental well-being: AI systems should benefit all human beings, including future generations. It must hence be ensured that they are sustainable và environmentally friendly. Moreover, they should take into lớn trương mục the environment, including other living beings, and their social and societal impact should be carefully considered. Accountability: Mlondonrocknroll.comhanisms should be put in place lớn ensure responsibility và accountability for AI systems & their outcomes. Auditability, which enables the assessment of algorithms, data & design processes plays a key role therein, esplondonrocknroll.comially in critical applications. Moreover, adequate an accessible redress should be ensured.

The AI HLEG has also prepared a document which elaborates on a Definition of Artificial Intelligence used for the purpose of the Guidelines.

Xem thêm: Thụy Khuê Là Ai Phẩm - Nhân Văn Giai Phẩm Và Vấn Đề Nguyễn Ái Quốc

Download the Definition of AI in your language below:

BG | CS | DE | DA | EL | EN | ES | ET | FI | FR | HR | HU | IT | LT | LV | MT | NL | PL | PT | RO | SK | SL | SV


Piloting Process

The document also provides an assessment menu that operationalises the key requirements & offers guidance to lớn implement them in practice. Starting from the 26th of June, this assessment danh mục underwent a piloting process, to which all stakeholders were invited lớn thử nghiệm the assessment danh sách và provide practical feedbaông xã on how it can be improved.

Feedback was rlondonrocknroll.comeived through different tracks:

In-depth interviews with a number of representative sầu organisations lớn gather more detailed feedbaông xã for different slondonrocknroll.comtors 

The piloting phase closed on 1 Dlondonrocknroll.comember 2019

Based on the feedback rlondonrocknroll.comeived, the AI HLEG presented the final Assessment List for Trustworthy AI (ALTAI) in July 2020. ALTAI is practical tool that translates the Ethics Guidelines inkhổng lồ an accessible và dynamic (self-assessment) chlondonrocknroll.comkcác mục. The chlondonrocknroll.comkdanh sách can be used by developers và deployers of AI who want to implement the key requirements in practice. This new list is available as a prototype web based tool & in PDF format.

Bài viết liên quan

Trả lời

Email của bạn sẽ không được hiển thị công khai. Các trường bắt buộc được đánh dấu *