The Future of AI: Conversing with Machines, Confronting Their Limitations

The Future of AI is a Conversation with a Computer

The Crux: AI writing tools like OpenAI’s GPT-3 offer a glimpse into the future of artificial intelligence, where machines can engage in fluid conversations and generate human-like text. While these models display impressive fluency, they also reveal significant limitations in truly understanding language and the world.

Case in point: The author used GPT-3 to generate text for an article on the future of AI writing. The model was able to produce plausible-sounding paragraphs, complete with fabricated quotes and references. However, closer inspection revealed that the content was often incoherent, contradictory, or simply made up.

Go deeper: Large language models like GPT-3 learn to generate text by mining vast datasets of online content. This allows them to mimic human language patterns, but they lack the embodied understanding and common sense that humans develop. As a result, they can confidently assert nonsensical claims when pushed beyond their training data.

Why it matters: The rise of AI writing tools has significant implications for the future of content creation. While they may excel at “low-attention” text like marketing copy or news briefs, their limitations raise concerns about the potential spread of misinformation and the erosion of human-crafted writing.

The big picture: Interacting with AI writing programs offers a glimpse of how we may engage with increasingly intelligent software in the years to come. Rather than passive recipients, users can participate in a dialogue, shaping the output through prompts and edits. This collaborative dynamic could lead to new forms of creative expression, but also raises questions about the extent to which these systems will influence and shape human thought and communication.

What they’re saying: “A deity that rules communication is an incorporeal linguistic power,” says artist and coder K Allado-McDowell, who explored this collaborative relationship with GPT-3 in the book “Phamarko AI.” The experience, they say, was akin to “communing with gods” and being “part of a larger ecosystem than just the individual human or the machine.”

The bottom line: As AI writing tools become more prevalent, they will force us to grapple with fundamental questions about the nature of intelligence, creativity, and our relationship with technology. While these systems may excel at certain tasks, their limitations also highlight the unique value of human-crafted communication and the importance of maintaining a thoughtful, critical dialogue with our artificial counterparts.