Regulate Artificial Intelligence
Regulate Artificial Intelligence
Technology is Progress, but not for everyone. Requesting a car ride or ordering takeaway food is now only a few keystrokes away, thanks to millions of workers who have to return to 19th-century style employment: no health care, no rest days, no sick leave, no pension and no employment protection.
Enters Artificial Intelligence (AI)... with the potential to disrupt our society with unprecedented impact and scale. Its danger is well documented: "AI algorithms embedded in digital and social technologies can encode societal biases, accelerate the spread of rumours and disinformation, amplify echo chambers of public opinion, hijack our attention, and even impair our mental wellbeing (1)." In a world of ever-rising inequalities, AI may be sealing forever the gender inequality, aversive racism and class divide.
This outcome is far from inevitable: the Tech industry is well aware of the problem and large IT companies are actively working on "ethical AI", free of bias and prejudice. As for regulation, the European Union (EU) is at the forefront thanks to its High-Level Expert Group on Artificial Intelligence, which has published in 2019 a series of guidelines to promote ethical AI. The new European Commission's president, Ms. Ursula von der Leyen, has expressed her interest for AI-focused legislation similar to the General Data Protection Regulation.
However, at the same time, the European Union has increased its annual investments in AI by 70% under the research and innovation programme Horizon 2020. It will reach EUR 1.5 billion for the period 2018-2020. With such a strong economic focus, the emphasis for the development of AI is surely on the industry. European Union officials already point to the fact that their ethical AI guidelines are meant only as a code-of-conduct for the industry, rather than enforced regulation. As always, some in the industry are reticent about such regulation: they welcome and actively participate in ethical AI, as long as it does not hinder their competitiveness and profitability.
So even though ethical AI is technically possible, the lack of enforced regulation would likely see the arrival of many AI services that amplify and perpetuate the prejudices and bias that currently exist in our society:
- medicine: AI will be used for automated diagnostic. Health data being mostly white and male (2), it could lead to more misdiagnoses among non-white and/or female patients
- education: AI will be used to identify students who show potential, or reversibly, who fall behind the rest of the class. If the design and training of such AI are not done with inclusivity in mind, it would set in stone the existing racial education divide
- police, military: facial recognition software linked to AI could institutionalise racial profiling
- financial: as of today loan and mortgage applications are often decided by AI. Studies (3) have shown that AI is already promoting racial discrimination in mortgage applications
- employment: in today's society, white and male candidates are favoured for hiring and promotion. Based on this raw data, an AI for hiring/promotion selection would perpetuate such bias
Leading tech experts have expressed their concerns but whose voice has ever reigned above the authoritative industry leaders on their self-serving mantra "progress for the people unhindered by archaic legal regulations"? With your voice, we must now demonstrate where we stand: our refusal to live in a society where only the most affluent and privileged ones benefit from the progress of science and technology, at the expense of the most vulnerable segment of the population.
It is AI that needs to adapt to what we hold dear as a society, not the other way round.
The European Commission is going to legislate on Artificial Intelligence. Sign this petition to ask that the legislation includes mandatory and enforceable rules for any AI reaching residents of the EU to be transparent, accountable and fair.
This petition is part of an art project called Daimon by Gina Peyran-Tan and Olivier Peyran.
1 Why We Need to Audit Algorithms, James Guszcza, Iyad Rahwan, Will Bible, Manuel Cebrian and Vic Katyal, Harvard Business Review
2 If AI is going to be the world’s doctor, it needs better textbooks, Dave Gershgorn, Quartz
3 Consumer-Lending Discrimination in the FinTech Era, Robert Bartlett et al., UC Berkeley