Treacherous Tech (Taking the Risk of Artificial Intelligence Seriously)

When experts first warned that artificial intelligence had a 10 to 20 per cent chance of causing human extinction within three decades, many found their claims alarmist. Now concerns about existential risks have intensified, even among developers. But only an informed, mobilized citizenry can compel governments to halt AI’s progress.

Treacherous Tech offers a framework for understanding why increasingly autonomous systems could threaten humanity’s survival. Beginning with the emergence of artificial intelligence, William Leiss and Richard Smith trace the collapse of safety guardrails as AI systems evolved from controllable tools into autonomous agents. Assessing the likelihood of known risks and the severity of their consequences, the authors show why institutions governing frontier AI cannot keep pace, failing to adequately address catastrophic perils such as cybersecurity threats and the creation of biological weapons. Leiss and Smith join Canadian parliamentarians in a call to prohibit the creation of superintelligent machines, urging Canada to work with other countries toward an international AI safety treaty.

For readers concerned about the stakes of artificial intelligence, this book offers a timely and pressing call to understand the risks – and to act while we can.