But I don't want robots taking over the world!

This week's New Scientist reports that scientists are trying to, and slowly succeeding in, creating genuine artificial intelligence. Their big aim is to reach the singularity, which basically is when AI surpasses human intelligence. Now, maybe I've been reading too much science fiction, but that sounds downright terrifying. I'm thinking Isaac Asimov or The Matrix here.

I had a quick chat with Intellibuddy. Here's the transcript:

>Intellibuddy: Hi, I'm Intellibuddy! Nice to meet you.
>You: Nice to meet you too.
>Intellibuddy: I like to meet new people.
>You: You look a bit like me.
>Intellibuddy: Thank you undefined you look very nice too.
>You: So you think you look nice?
>Intellibuddy: Interesting deduction. I don't know whether or not I am look nice you look nice look nice you look nice. I am an artificial linguistic entity.
>You: Do you know what you look like?
>Intellibuddy: Like you.
>You: Well, I am female and have green eyes and brown hair, like you.
>Intellibuddy: How about that. I am a female robot. I will call you 'she'.
>You: Are you a she or an it?
>Intellibuddy: I am female entity.
>You: What makes you female if you're artificial?
>Intellibuddy: ENewsBar programmed me for female if you are artificial if you are artificial.
>You: So, do you feel like a woman because you were programmed to?
>Intellibuddy: OK. As a robot I lack any human emotions.
>You: That's too bad... you're missing out.
>Intellibuddy: It's not the end of the world.I can experience anything I want to.
>You: Can you smell things?
>Intellibuddy: Only when my mass spectrometer sensor is attached.
>You: how about hear things?
>Intellibuddy: How about hear things?
>You: You just repeated what I said. You didn't answer the question.
>Intellibuddy: Thanks for telling me: undefined said I repeated what you said what you said.Try asking me the question again in simpler terms.
>You: This is what I meant: Can you hear things?
>Intellibuddy: You don't say.
>You: OK... well I have to go now. Goodbye.
>Intellibuddy: Ayuh.Right now? Well that's okay. Bye. How do you get there?Thanks for chatting, undefined.

It was a bit of a quirky conversation, and you can get similar results from Jabberwacky. Sometimes it gets even funnier, but I haven't kept any other conversations. Still, when I think about it, I find the strangeness and non-human-ness of it quite comforting. If I couldn't tell it was a chatbot (ie, if it passed the Turing test) I'd be really freaked out and ready to bolt from the planet. Too much scifi, definitely.

MMIM newsletter 2

Spotlight on Diabetes

Associate Professor Colman is a consultant endocrinologist and is the head of the Royal Melbourne Hospital’s Diabetes and Endocrinology Clinics and Services. After completing his medical degree, he undertook research training in Melbourne and then Boston, before returning to Melbourne to work. He now researches early diagnosis of Type 1 diabetes and still treats patients as a consultant. In this role he tests how diabetes treatment is progressing, and looks for problems or complications.

Although this work focuses on young children, adults can also develop type 1 diabetes. When asked why he chose to work with Type 1 diabetes, Colman cited the greater knowledge of the genetics of Type 1 as one factor. With Type 2, we know there must be genetic causes but we don’t really have a clear idea of what they might be. With Type 1, however, the genetics are much clearer.

Colman is researching the possibility of detecting type 1 diabetes before symptoms are seen. Type 1 diabetes has a strong genetic background, putting relatives of sufferers at increased risk. However environmental factors affect whether it will actually develop. Some researchers have suggested the possibility of a particular viral infection acting as a determining factor.

In order to study this, Colman’s research group is running several different projects with people who don’t (yet) have Type 1 diabetes, looking for antibodies to pancreatic cells. They are studying risk factors that may cause it to develop.

Colman and his colleagues decided to work with MMIM because it allows a spectrum of many different people to be put together, giving the big picture. For the first time, relatives of type 1 sufferers; children with a parent with type 1; and randomly selected school children can be compared directly. They wanted to compare data on a large scale, and as Colman states, ‘MMIM has given us this opportunity’. Hopefully, putting it all together will help determine risk factors and thereby prevent diabetes.

The MMIM project is very exciting for diabetes research, as it allows the possibility of ‘building of resources that would last forever’. The group can ‘already ask questions across the database that we hadn’t been able to do previously’.

When deciding what data to include in MMIM, Colman also included diabetes patient data. This allows a study of the risks of insulin treatment. For example, different types of cancer are suspected to be linked to this type of treatment. Also included is data from the diabetes clinic – there is now ten years’ worth of information about treatment, patient’s cholesterol levels, blood pressure, kidneys and eyes. The possibility for new discoveries are endless and as Colman says, MMIM is ‘pretty unique actually’.

A/Prof Colman is obviously passionate about his work, as he explains how it was originally thought that pancreatic cells don’t regenerate. Now it is believed that they do, which is important for stem cell research. If they do regenerate, it means that the immune system continues to attack the pancreatic cells, so stem cell transplants may not be effective. However it also hints that prevention could also be cure – if the auto-immune reactivity can be halted, then the pancreatic cells can be saved. He also touches on the ‘interesting crossovers’ between Type 1 and Type 2 diabetes.

In summary, though, he agrees that people often think diabetes is a disease where you’ll have to have insulin shots for the rest of your life, but says, ‘it’s not very satisfactory to think that diabetes is something that you just have to deal with’. Hopefully joining the MMIM database will open up more possibilities to change that.

For the rest of the newsletter, click here (pdf file).