Alan Turing was many things: a wartime hero, a computer scientist who pre-dated computers, the father of artificial intelligence and a persecuted gay man. Now, 70 years after his death, he will also become a museum exhibit in chatbot form.
Bletchley Park, the site where he worked as a code breaker during the second world war, is working with a UK company called 1956 Individuals to create an AI model that can 鈥渃onverse naturally with visitors鈥� as Turing himself. The idea is to tell his story in an interactive way that grabs audiences. But is such a project ethical, respectful or even useful?
Advertisement
Some late public figures have estates managing their image and works: Albert Einstein bequeathed the rights to his 鈥渓iterary property鈥� to the Hebrew University of Jerusalem, which now benefits when he appears in advertising. In one ad, he promotes smart energy meters from the bath. Einstein had a sense of humour, though whether he would appreciate that particular ad can never be known. He chose to grant permission for his works to pass into someone else鈥檚 control (the concept of public image rights didn鈥檛 exist at the time). There is no indication Turing did this.
Dermot Turing, Alan Turing鈥檚 nephew, has backed the project. He wasn鈥檛 available for interview, but we must ask, as a society, if we are happy to let this happen without the subject鈥檚 direct consent. Similar concerns arose when Authentic Brand Group, which owns the public image rights for several late celebrities, allowed an AI .
These are thorny issues and they will continue to arise, even for those of us who aren鈥檛 famous. Projects to recreate grieving relatives鈥� loved ones from troves of their emails and texts have led to some experts suggesting that legislation is needed to prevent people being digitally reincarnated without their prior, living approval.
There is also the question of whether current AI can do Turing justice. We don鈥檛 know how much of him will actually reside in this neural network or whether his surviving writing will be enough to train it to match his tone and opinions. Nor do we know how much of his complex story the model will be willing to address, or how sensitively it can do so.
Turing was charged with 鈥済ross indecency鈥� by the state simply for being in a relationship with another man. When convicted, he was forced to choose between imprisonment or chemical castration. He was also stripped of his security clearance, a decision through which the UK foolishly deprived itself of Turing鈥檚 genius. He eventually took his own life. These are painful experiences that current AI models are incapable of tackling sensitively. Building this Pandora鈥檚 box is ethically shaky and opening it to public display is reckless 鈥� the inevitable problems can be seen a mile off.
What would Bletchley Park鈥檚 curators do if the model lashes out at those who did him harm 鈥� or, worse, condones them? What would AI-Turing have to say about Russia鈥檚 invasion of Ukraine or the upcoming US election? No doubt safeguards would be put in place to prevent the model straying into controversial territory where his opinion can only be speculation, but then what purpose does a sanitised, gagged Turing serve in helping us understand the man?
The creators of this project tell me that they haven鈥檛 yet built the model, but will be able to 鈥渃arefully control the responses with pre-agreed answers, when triggered by certain keywords鈥� so responses will never be inappropriate, steering away from opinions towards facts. But in a world where AI created by Google told people to , what hope does Bletchley Park have of navigating this without incident?
Turing鈥檚 legacy should be remembered. But the man himself 鈥� recreations included 鈥� should be left to rest in peace.
Matthew Sparkes is a technology reporter at New 杏吧原创
