Al-Amn Magazine

childhood friend who was at your brother’s wedding, your partner who held your hand at the doctor’s office. Tests of AI empathy typically don’t compare a chatbot’s cold comfort with the kind of socially embedded care that truly nourishes us. If they did, the chatbots would lose. Or consider a recent paper claiming that AI can produce more novel ideas than human experts. In this case, the researchers asked computer science PhD students and postdocs to compete with a modified version of Anthropic’s Claude AI model. Their task: to come up with new research ideas. Each competitor was assigned a research topic (such as how to reduce bias in AI systems) and a template for writing out their ideas (to mask stylistic differences between human and AI writing). All the ideas were then judged by another group of PhD students and postdocs, who didn’t know whether the ideas came from AI or humans. The judges rated the AI-generated ideas as more novel, and in some cases more exciting, than the human- generated ideas. This seems impressive until you consider how scientific research actually works. Although stereotypes of scientists as lone geniuses persist, there is overwhelming evidence that science is a team effort and more diverse teams do more creative and impactful science. We’re most innovative when we draw on communities of knowledge, thinking together through complex problems. But that’s not the test that the researchers chose to pit humans against machines. The humans in the study had to come up with their ideas alone. Once again, the AI victory rings hollow when we consider that the humans in this game had to leave their best talents behind – in this case, their collaborative relationships. The same is true of a recent paper claiming that AI outperforms humans in resolving political disagreements. Researchers at Google invited small groups of British citizens to privately share their opinions on divisive issues like immigration and Brexit. The team built an AI system to act as a mediator, generating “group opinion statements” designed to maximize consensus among group members. They also trained human volunteers to perform the same task. The citizen groups reached more consensus when guided by AI mediators than human ones, rating the AI- generated statements as more clear, informative and fair than their human counterparts. Commentators suggested AI might hold promise for bridging our deepest political divides. But a closer look reveals how the study’s methods disadvantaged the human mediators. All communication took place over an online chat interface. Anyone who’s ever tried to resolve a disagreement by texting knows how much crucial information vanishes across a digital divide. Effective mediation involves attuning not just to what people say, but how they say it: their tone of voice, facial expressions and body language are vital cues that help us distinguish grudging acceptance from hearty agreement. AI mediators might have an edge in online discourse that excludes these cues, but that’s no reason to assume they’re superior tools for healing our political rifts – not least because online discourse contributed to those rifts in the first place. Considering why these tests are unfair helps us better appreciate our distinctively human talents. We build relationships thick with meaning and forge new knowledge out of trusting social bonds. We find common ground reflected in our faces, voices and postures, perhaps because embodied communication helps us recognize our shared mortality. No chatbot can achieve these feats, because chatbots don’t have bodies, and aren’t embedded in our social worlds the way our loved ones are. Techno-optimism is more accurately described as “human pessimism” when it assumes that the quality of our character is easily reducible to code. We can acknowledge AI as a technical achievement without mistaking its narrow abilities for the richer qualities we treasure in each other. This is what we›re up against Bad actors spreading disinformation online to fuel intolerance and undermine democracy. A media ecosystem dominated by a handful of billionaire owners. Teams of lawyers from the rich and powerful trying to stop us publishing stories they don’t want you to see.  Lobby groups with opaque funding who are determined to undermine facts about the climate emergency and other established science.  Authoritarian states with no regard for the freedom of the press. theguardian.com