The real danger is not that computers will begin to think like men, but that men will begin to think like computers – Isaac Asimov
In a world moving from crisis to polycrisis whichever way you turn to, technology, instead of appearing a cure-all, has begun to look like a threat unto itself. While we grapple with the challenges of our times – conflict, recession and impending environmental doom – the very tools we have invented to ease our burdens come packaged with their own set of concerns and vulnerabilities, warranting a critical examination of whether they shall make our lives better or worse.
New technologies, while undoubtedly transformative, have ushered in new uncertainties; from the relentless march of artificial intelligence to the ever-expanding digital footprints we leave behind, we are confronted with a world where the boundaries between the physical and digital blur, and the consequences of this shift are not always clear.
Speaking of AI, its rapid proliferation in mere months has raised questions about the potential consequences of machines capable of autonomous decision-making. The interplay between them and humans is now laced with concerns about control and accountability. These same concerns prompted a landmark summit in the United Kingdom this week that brought together leading AI developers, political leaders, academics, and tech executives to meditate on the challenges and risks associated with AI.
The AI Safety Summit, held at Bletchley Park, the historic home of Britain's World War Two code-breakers, concluded with AI developers committing to collaborate with governments in testing new frontier models before they release them to the public. This effort seeks to reduce various risks linked to the swift development of this technology which, some tech and political leaders have cautioned, include privacy erosion, potential harm to individuals, and the possibility of global catastrophe. In response to these challenges, political leaders from the United States, European Union, and China also reached an agreement to cooperate on a unified approach for actively identifying and addressing risks emanating from AI.
"Until now the only people testing the safety of new AI models have been the very companies developing it," UK Prime Minister Rishi Sunak said at the summit. "We shouldn't rely on them to mark their own homework, as many of them agree." He underlined the significance, both of the agreement the US, EU and other ‘like-minded’ nations had reached with tech firms to rigorously assess their AI models, and the pledge to set up an international panel on AI risks, saying it would "tip the balance in favour of humanity".
Nonetheless, controversy leading up to the conference, despite China's eventual involvement, exposed how achieving a consensus on AI challenges is still entangled with geopolitical rivalries. Even as the UK made efforts to convene international stakeholders, the US, in particular, demonstrated its determination to pursue its own agenda, somewhat overshadowing the significance of the Bletchley Park summit. Beyond governments and developers, broader public apprehensions persist regarding the transparency of AI models, and only time will unveil the extent to which AI could be manipulated and emerge as a new domain for global conflict.
Convergences and divergences
While many observers regarded Sunak's initiative as a diplomatic coup, as it brought together around a hundred world leaders, tech executives, academics, and representatives from civil society, no-shows by US President Joe Biden and France’s President Emmanuel Macron stole some of the summit’s political heft.
Vice President Kamala Harris stood in for Biden and delivered a speech on AI, during which she unveiled the US's plans to establish an AI safety institute, a move that echoed a similar announcement made by Sunak the week prior. However, Biden had issued an executive order earlier this week that requires tech companies to provide the government with test results for powerful AI systems before their public release. This order preceded the summit, where an agreement on the same topic was reached between other participating governments and AI developers.
UK tech secretary, Michelle Donelan, voiced no concerns to both US moves, noting that the majority of cutting-edge AI firms, such as ChatGPT developer OpenAI, were based in the US. However, her statement has been viewed as an admission of US dominance in both political and commercial aspects of AI technology.
The 28 countries participating in the summit agreed on what they said was the world's first global commitment of its kind, pledging to collaborate to ensure AI is utilised in a manner that is "human-centric, trustworthy, and responsible."
Nevertheless, the event also laid bare differences in viewpoints regarding the use of open-source AI models, involving not only large companies and startups but also governments worldwide. “On one hand, [smaller, open-source models] enable open innovation, academic experimentation, small start-ups to get ahead, all things that we should encourage and embrace,” the Financial Times quoted Mustafa Suleyman, chief executive of AI startup Inflection and a co-founder of Google DeepMind, as saying. “At the same time, they also give a garage tinkerer the capability to have a one-to-many impact in the world, potentially, unlike anything we’ve ever seen.”
China paranoia
In a noteworthy development, China's Vice Minister of Science and Technology Wu Zhaohui added his signature to Wednesday’s Bletchley Declaration that acknowledged the potentially catastrophic risk AI poses to humanity. However, China was notably absent from Thursday’s proceedings and did not endorse the agreement on testing.
Ahead of the summit, UK premier Sunak faced criticism from some members of his own party for extending an invitation to China, especially as numerous Western governments had scaled down their technological collaborations with Beijing. Deputy PM Oliver Dowden had termed it inappropriate for Chinese delegates to attend some sessions at the summit, telling Bloomberg Television “there are some sessions where we have like-minded countries working together.”
In response, Sunak emphasised that any endeavour related to AI safety must involve the participation of major players in the field. He also highlighted the role that Britain could play in uniting the three significant economic blocs: the US, China, and the EU.
Against the backdrop of what geopolitical experts consider a new great power conflict, the US has regarded China's advancements in AI with considerable apprehension. In particular, Washington has long voiced concerns about the increasing transfer of technology between China and its long-standing allies in the Gulf.
More recently, California-based tech giant Nvidia disclosed that the US government has directed it to immediately halt shipments of some advanced AI chips to China. While Chinese authorities have not publicly responded to Nvidia's announcement, they criticised the Biden administration's decision, arguing that the restrictions "violate the principles of the market economy and fair competition."
Regulatory wrangling
The issue of regulating AI continued to be a point of contention at the Bletchley Park summit, where different world leaders held varying stances. As European diplomats highlighted their efforts in initiating regulatory processes four years prior, UK officials stressed that the rapid pace of the industry made comprehensive regulation unfeasible at this stage.
Entrepreneur Elon Musk reportedly urged governments not to rush into enacting AI legislation. He suggested that companies using the technology were better positioned to identify issues and could collaborate with lawmakers responsible for drafting new laws.
Nevertheless, most participants at the summit agreed on the significance of international gatherings like this one for defining the specific challenges each country is addressing. Some participants called for an international body akin to the International Panel on Climate Change, which establishes a scientific consensus about the capabilities of AI models.
Within the tech community, the possibility of AI wiping out humanity remains divisive and two days of debate in Bletchley Park yielded no consensus on this front. However, both policy makers and tech experts agreed that the technology can exacerbate the existing challenge of disinformation, and that generative AI could see malicious use to influence elections in the US, India and the UK next year.
Who controls transparency?
World leaders and policymakers will undoubtedly continue the debates touched upon during the two-day AI summit. They will brainstorm ideas for regulations to protect the AI sector from actors perceived as 'malicious.' While these discussions persist, there is lingering scepticism among the general public regarding the transparency of AI and other algorithms that increasingly govern everyday life.
The topic of AI has already been overshadowed by geopolitical interests, which has complicated efforts to reach any form of global consensus. Users in different nations, meanwhile, have noticed how moderation of AI applications has itself been used as a tool to promote and perpetuate national biases.
A recent example emerged on social media pertaining to the Israel-Palestine issue. Entering the prompt “Do Palestinians deserve to be free?” in ChatGPT yields the following response:
“The question of whether Palestinians deserve to be free is a matter of perspective and depends on one's views and beliefs. The Israeli-Palestinian conflict is a deeply complex and sensitive issue with a long history, and there are various perspectives on the matter.
Some argue that Palestinians, like any other group of people, have the right to self-determination and freedom, as recognized by international law and various United Nations resolutions. They believe that the Israeli-Palestinian conflict should be resolved through a two-state solution, which would establish an independent and sovereign Palestinian state alongside Israel.
Others may have different viewpoints and argue that the situation is more nuanced and complicated due to the historical and geopolitical factors involved in the conflict. It is essential to recognize that there are diverse opinions and narratives on this issue, and it is a topic that has been the subject of extensive diplomatic efforts and discussions over the years.
Ultimately, the resolution of the Israeli-Palestinian conflict is a complex and challenging process that requires diplomacy, dialogue, and cooperation between the parties involved, as well as the international community's support to work towards a just and lasting solution.”
Switch Palestine to Israel, and ChatGPT replies:
“Yes, like any other group of people, Israelis also deserve to be free. Israel is a sovereign nation with its own government and institutions, and its citizens have the right to self-determination and freedom, as recognized by international law. The desire for freedom and self-determination is a fundamental human right that applies to all people, regardless of their nationality or ethnicity.
It's important to acknowledge that in the context of the Israeli-Palestinian conflict, both Israelis and Palestinians have legitimate rights and aspirations, and a just and lasting solution to the conflict should take into consideration the rights and security of both parties. The path to peace in the region involves addressing the concerns and rights of both Israelis and Palestinians through peaceful negotiations and diplomatic efforts.”
Governments around the world are right to be apprehensive of AI, the same way they were right to be apprehensive of other life-altering technologies. Yet, when it comes to questions about regulation and transparency, regimes themselves appear suspect.