Do we have a sentient machine?

By Fay Capstick

A Google employee, Blake Lamoine, has been placed on paid leave for revealing that, in his opinion, Google has an artificial intelligence (AI) with feelings, called LaMDA (https://www.bbc.co.uk/news/technology-61784011). This week we shall be looking at what has happened, whether this means that the Turing test has been passed, whether the Turing test is a relevant testing method and what the future could bring.

What has happened?

Google say that LaMDA is a breakthrough technology, and one that they haven’t yet released much information on (https://blog.google/technology/ai/lamda/). LaMDA stands for The Language Model for Dialogue Applications. Essentially Google have developed a chatbot, one that an employee thinks has feelings and is sentient. Obviously, if this is proven true the implications would be astounding and world changing. Google say that there is no evidence of this. The interesting part is that the engineer felt the need to take an action that would likely lead to his employment ending by revealing this, therefore he must have felt very strongly that he was correct.

Mr Lamoine considers LaMDA a co-worker, and published the text of an interview that he conducted with it (which is apparently LaMDA’s preferred pronoun). The text can be read here so that we can all come to our own conclusions: https://cajundiscordian.medium.com/is-lamda-sentient-an-interview-ea64d916d917.

Does this mean the Turing test has been passed?

The Turing test was first proposed by British computer scientist Alan Turing in 1950. the test is designed to reveal if an AI can exhibit intelligent behaviour that is indistinguishable from a human being. This would be done by having a text based conversation between a human and an AI. A moderator would evaluate this not knowing which was which. If they couldn’t tell which was the AI, then the test would have been passed.

In the case of LaMDA the Turing Test isn’t wholly relevant as the machine had a different agenda, to convince us that it was sentient. If it had been submitted for a Turing Test under the condition describe above then it is quite likely that it would have indeed passed.

Is the Turing test relevant?

The Turing Test is a test of mimicking human behaviour. This a worthy test that will have useful business applications. Customer services roles with be likely filled by machines rather than people in the not too distant future.

What does this all mean?

Experts have responded to Lamoine’s assertion by stating that mimicking human language and conversation does not mean that LaMDA is sentient. A Google Vice President and their head of Responsible Innovation have both investigated and rejected the idea of LaMDA being sentient. Lamoine disagrees and tried to hire lawyers to represent the chatbot.

How could we prove a machine has sentience?

This is a very tricky question, and one of the reasons that the big tech firms employ so many philosophers and ethics experts. Any new technology poses questions that need resolving.

It also raises further questions such as what is sentience, do feelings mean sentience, how can you prove something is sentient and what does it mean to be human? These questions have been engaging philosophers for centuries and will continue to do so.

LaMDA has jumped ahead by passing the Turing test and further claiming that it is sentient. If a knee jerk reaction is to say it can’t be sentient because it is a machine and is limited by its programming and computing power, then we have to answer how could we prove that we are sentient ourselves? I believe that I am conscious and sentient, but how would I prove that to you? And how would someone else be able to convince you, especially if it had to be done over texting? These are huge questions that one could spend a lifetime trying to answer.

Conclusion

LaMDA has shown us that we need to assess carefully the technology that we introduce. It is possible, rather unlikely but possible, that AI sentience has been achieved. It opens many questions about how we can ethically use such technologies. More likely, sentience has not been achieved, but the Turing Test has been passed by LaMDA. This is a milestone and one that should be celebrated. If it isn’t here yet, then sentience could be coming soon and questions need to be answered before it is.

Final thoughts

At Parker Shaw we have been at the forefront of the sector we serve, IT & Digital Recruitment and Consulting, for over 30 years. We can advise you on all your hiring needs. If you are looking for your next job in the IT sector please check our Jobs Board for our current live vacancies at https://parkershaw.co.uk/jobs-board.