WHY TECH COULD TEAR US APART AGAIN
Geoffrey Hinton is the so-called ‘Godfather of AI’ and his view of the risks of ‘AI getting super smart and understanding it doesn’t need us’ has led to lots of talk about an AI apocalypse where AI doesn’t just come for your job, but like Arnie’s Terminator, it comes for your clothes and your life, too. This existential risk can’t be discounted, but the impact of AI already on how we understand the world and relate to one another is a more immediate problem.
Like many, I have long thought that the defining challenge of tech in our era is how social media rewires us individually and in community. Not anymore. It is just preparing the way for what AI will do: John the Baptist to Jesus. Such is the capacity of AI to explain and to problem solve in seconds, some will come to depend on it with the same trust levels reserved for Jesus, too – and pay a price.
All technological advances emerge as a mixed blessing. The capacity of AI to do good is phenomenal, but the same issues we face with social media will become more pronounced with AI. Tech companies are profit driven, indisposed to code in safeguards and keen to distract us from anything other than being online and on their platform.
We were too naïve in our embrace of social media, assuming it would only bring us together when algorithms were subversively designed to make us angry with other people because, as Silicon Valley has it: enragement = engagement. With AI, the means are different but the ends are the same.
Large Language Models (LLMs) like ChatGPT and Gemini are trained on billions of words and use these to craft responses to the questions we pose. These billions of words are found online, and so minority professional judgements can be drowned out by a cacophony of opinions expressed by people who think they know stuff but don’t; who disbelieve the truth; whose default mode is attack. It’s a bit like recycled air on a plane: on one level a miracle of engineering at 40,00 feet, but passengers are breathing contaminated air that is bad for them.
Greater still is the risk posed by how we ask questions of AI in the privacy of our homes. Few people will bother to frame their questions in as unbiased a way as possible, not least because we are unaware of our biases; yet how we ask those questions will determine the kind of answer AI gives us. This introduces the risk that LLMs will become the greatest source of confirmation bias in the world: feeding our prejudices back to us and offering us more content to reinforce our discrimination. It mimics social media in delivering us an echo chamber where we are surrounded by information that confirms our opinions rather than challenges them, except the information won’t be delivered by a human Reddit chain, but by an apparently incorruptible chatbot.
To use AI well compels us to be better at how we ask our questions; to reactivate the method of enquiry pioneered by Socrates and seen vividly in the scriptures where the questioning of elders introduces people to the story of salvation: ‘what do these stones mean?’ (Joshua 4: 21). It asks for a dose of humility appealed to in the parables of the Pharisee and the sinner (Luke 18: 9-14). If we approach AI with Pharisaic self-satisfaction and the certainty of our standing, it will reinforce this. Far better to be the person who is not wise in their own eyes and has self-awareness in how they learn (Proverbs 3:7).
Software engineers work all the time to iron out glitches in technology, including round the biases learned by LLMs in training, but they will not remedy every problem and inevitably bring their own flaws to bear in design. Theo Baker wrote ‘How To Rule The World’, about his experiences as an undergraduate at Stanford University, which is adjacent to, and the tech nursery for, Silicon Valley. When he asked students in his IT class who ‘thought ethics were important to their work’, not one raised their hand. It sounds like it would be foolish to rely on the next generation of tech wizards to save us, too.
Jesus said the truth would set us free (John 8:32), and truth has become the key battleground of this developing century. In this environment, how we ask our questions will become a vital characteristic of the wisdom that comes from above, which is ‘without a trace of partiality’ (James 3: 17).
