Is Artificial Intelligence going to be good or bad for society? - Bill Carmichael

And hold onto your hats folks because this one could be a massive problem that some authorities say could wipe out human life on the planet.

Dozens of experts, including many leading figures in the industry, signed a statement this week warning Artificial Intelligence, normally abbreviated to AI, could lead to the extinction of humanity, and they compared the risks to those posed by global pandemics and nuclear war. The statement, published on the website of the Centre for AI Safety, warned that AI could be weaponised, for example drug discovery machines could be used to build new chemical weapons.

The immense power of AI could also be concentrated in a few hands and authoritarian regimes could use it “to enforce narrow values through pervasive surveillance and oppressive censorship”.

Chilling stuff, but what exactly is AI? The first thing to note is that Artificial Intelligence isn’t very intelligent at all, in the sense we apply that word to humans, and can sometimes be remarkably stupid. Instead computers use vast amounts of data to detect patterns and make decisions.

For example, a building society might feed in thousands of past mortgage applications, both successful ones and those that have been turned down, into a computer. By looking at this data the computer “learns” what a successful application looks like, and then uses this criteria to approve or turn down future applications, without any human involvement.

But sometimes things can go disastrously wrong. For example, a few years back the giant mail order company Amazon started using AI to recruit software engineers and other technical staff by feeding into a computer the CVs of thousands of past employees.

The problem was that in the past its technical staff was overwhelmingly male, and the machine taught itself that male applicants were preferable, and it began rejecting female applicants no matter how highly qualified they were.

There are also fears that AI will replace many thousands of jobs, and not just low skilled ones. Areas such as accountancy, administration and journalism (gulp!) are at risk.

Personally, after another nightmare commute by car this week because of yet another rail strike, I’d be more than happy for robots to replace train drivers who are paid £60,000 a year to drive in a straight line. That’s one job a computer could do better and more cheaply than a human.

In higher education there is increasing evidence that students are using AI to produce work for assessment. There is, for example, a generative AI system called ChatGPT that is entirely free to use. Students can ask the machine to produce, say for example, a 3,000 word undergraduate essay on a given topic.

The result is pretty much undetectable, and even if the tutors suspect the student cheated by getting a machine to do the work for them, there is no way to prove it.

I expect in the future there will be fewer essay assignments, and more traditional exams or timed assessments using computers with no internet access, and face to face vivas where students are questioned about what they have learned - all to make sure it is the student who has done the work, not a machine.

But despite all the problems we shouldn’t forget that AI can also be a tremendous force for good, making processes faster and more efficient, finding new drugs to fight cancer and making digital learning better for children.

For example, recently a Swiss-French team of researchers helped a Dutch man paralysed for 12 years after a cycling accident, to walk again using AI. It is good that concerns are being raised now, so we can assess the risks and put in place procedures to keep us safe.

As Mira Murati, the Chief Technology Officer at OpenAI, which developed the ChatGPT system, said: “AI systems are becoming a part of everyday life. The key is to ensure that these machines are aligned with human intentions and values.”