AI - Artificial Intelligence: boon or bollocks?

When faced with a chatbot, do you:

  • Close the chatbot and continue using the site?

    Votes: 9 29.0%
  • Close the website and go elsewhere?

    Votes: 5 16.1%
  • Engage with the chatbot and use it to answer your questions?

    Votes: 0 0.0%
  • Ask the chatbot what underwear (if any) it is wearing and would it like to engage in procreation?

    Votes: 7 22.6%
  • Give me a human to talk to or I'll burn down your head office!

    Votes: 10 32.3%

  • Total voters
    31
#21
Some of the Chatbots in use are quite primitive, and really only respond using fuzzy logic to match questions against known answers, so are more of a glorified search engine with an avatar than really AI
Most are, as you say, fuzzy logic. Other applications passed of as AI tend to be expert systems which are basically glorified decision tree's, or computerised flow charts, ie medical diagnosing systems. The thing with the medical diagnostic system is that you can generally program in more 'text book' knowledge that a Dr. can remember off the top of his head. Where the medical systems fall down is that they cannot assess your skin colour, pupil dilation and general demeanour - though last time I was looking at them they were rated as better than a human Doc for diagnosis from questioning alone.

I used, and even wrote, some Back Propogation Networks and Kohonen Self Organising Maps when doing my research studentship. I used the BP Networks to search large databases for patterns and to predict probability, the K SOM's were used almost like a human iris to detect patterns in visually presented media such as maps. When messing around I also used them for stock market prediction, house valuation, predicting horse race winners, finger print recognition and retail location modelling.

The thing with AI is that it needs to be cleverly designed, not necessarily intellectually cleverly, more sort of imaginatively. I used to have a knack of combing different data-sets imaginatively which the career academics just could not do. I put it down to having kicked around in the army, coppering, doing IT related degrees before moving out of a uni school of IT into a uni school of geography. All the geographers had problems thinking outside the geography box. My mate, Big Steve, who I did my degree with was an ex-Met copper, after the degree he got a job working in a hospital research department. One day the Medical Dr. Professor he is working for say's, "it'd be marvellous if some could invent a computer program to do 'x'". Over his lunch break Big Steve used a piece of shareware software to knock out an expert system to fulfill his boss's wish. To me and Steve it was basic 1st year degree programming but, to a non-IT/non-programmer person it was black magic. The Prof had Steve put on a fully paid part-time PhD and Steve was happily banging out software for the department like there was no tomorrow.

The best designed piece of AI I was involved in was with a bank for predicting market movement of financial instruments and currencys. There was a team that included programmers, systems designers, historians, sociologists, bankers and a couple of others. When considering the movement of shares you have to factor in: Which party the PM is from, which party is in power, how long till the next election, are there any bye elections, what is the GBP vs. $ rate, did it rain today, what is the temperature, are we at war, has there been a terrorist incident, etc., etc. There was a system for each product and each system had over 300 inputs of data in order to spit out a result that equated to predicting either: Price going up, or price going down. That project took a couple of years. You cannot just have Eric the computer bloke knocking out systems that people have to interact with you need breadth of expertise - and these systems are despised and hated because Eric knocks them out with his team of computer nerds who have problems chatting/interacting with real people.
 
#22
Against all my better judgement, I would like to know your opinion about the use of chatbots/'AI' and other such crap when using websites.

Do you hate them, do you like them, or do you not give a toss? If you like them, please take a moment to say why. Have you come across any decent ones?

Why am I asking this? Because I work in insurance and it appears that insurance companies and brokers are utilising chatbots and what they call 'AI' at an ever increasing rate. When asked, they say it is because it increases customer satisfaction and engagement... I suspect it is more about sacking as many people as possible and replacing them with a piece of shite software and further commoditising insurance (something vehemently denied by every insurer).

I have seen AI used to great effect in medicine: being much more accurate and reliable than the best doctors at identifying cancerous cells, and better predicting the side-effects of new medicines (as just two examples). I have seen it used with zero beneficial effect by companies such as Amazon when trying to sell you crap similar to the other crap you may have already purchased, and its utterly counter-productive use by companies such as Adzuna and other jobsites for matching CVs and jobs (who knew that my ideal job was a lion-taming actuary?).
So as people who know me in real life, I;m doing a PHD in AI. I do the more mathematical parts of AI, applying it to medicine.

So regarding Chatbots. This comes under an area known as Natural Language Processing. This as developed very rapidly in the last 5 or 6 years mainly by 'deep learning',which in turn has developed massively in the last 10 years. The huge amount of interest in it means there's a significant amount of open source software available for organisations to build their own chatbots. Deep learning is the overwhelmingly successful approach to this and therefore most effort has been devoted to this and most remarkably its success is demonstrated across most languages.

Regarding amazon and product recommendations. It is no surprise that this is more varied in terms of success. Recommender systems come under an area known as 'collobrative filtering'. There a significant number of approaches. Unlike chatbots there are no overwhelmingly successful approach. Performance of different approaches is very much dependent on the field/domain and models are rarely scaleable i.e. a recommender system algorithm that works for TESCO may end up failing miserably for Netflix.

In terms fo the future, I actually think applications to medicine will eventually slow down due to ethical reasons. One of the big problems with AI methods is that they are not very interpretable. Its very difficult to understand why they work and when they fail. Obviously in the healthcare field this brings alot of ethical issues. Indeed, this is one of my areas of research. However for stuff like chatbots and Recommender systems, intpretability is less of an issue.
 
#23
With a very limited knowledge of computing, every time I hear or read about AI, I just don't believe in it. To me it seems that if you interact with a computer and it gives you information, it's a souped-up version of Conditional Formatting in Excel - you input something and the computer is programmed to give you a response against what you've said or typed.

That isn't AI, it's simply clever computer programming. Am I right or have I got the wrong end of the stick?
 
#24
Google AI is rubbish - whenever I ask it stuff it tells me it can't understand or gives me something totally different to what I asked. As google is basically one of the biggest leaders in the world on AI - I'd say other companies have still got a long way to go, but it will get there eventually.
 
#25
Its liner processing. You ask one question and if you are lucky you will get an answer to that question. They are very unsophisticated.

Yes - much more scary stuff is coming. I work in the trade.
To an extent a chatbot reminds me of the old apple scam about style over substance. The boxes and appearance is ever more pretty and attractive as times go by, but the actual service delivery and application is not great and whatever savings you make in implementation, has to be balanced to how much you want to irritate your customers.

But, the real deal is all focused on facial recognition especially with regards gauging emotions and the one thing I never dare ask my managers are the legal side. If in future world, a mistake is made and computers always make mistakes. GDPR and the odds of a massive rise in lawsuits, seem inevitable .
 
#26
So as people who know me in real life, I;m doing a PHD in AI. I do the more mathematical parts of AI, applying it to medicine.

So regarding Chatbots. This comes under an area known as Natural Language Processing. This as developed very rapidly in the last 5 or 6 years mainly by 'deep learning',which in turn has developed massively in the last 10 years. The huge amount of interest in it means there's a significant amount of open source software available for organisations to build their own chatbots. Deep learning is the overwhelmingly successful approach to this and therefore most effort has been devoted to this and most remarkably its success is demonstrated across most languages.

Regarding amazon and product recommendations. It is no surprise that this is more varied in terms of success. Recommender systems come under an area known as 'collobrative filtering'. There a significant number of approaches. Unlike chatbots there are no overwhelmingly successful approach. Performance of different approaches is very much dependent on the field/domain and models are rarely scaleable i.e. a recommender system algorithm that works for TESCO may end up failing miserably for Netflix.

In terms fo the future, I actually think applications to medicine will eventually slow down due to ethical reasons. One of the big problems with AI methods is that they are not very interpretable. Its very difficult to understand why they work and when they fail. Obviously in the healthcare field this brings alot of ethical issues. Indeed, this is one of my areas of research. However for stuff like chatbots and Recommender systems, intpretability is less of an issue.
You are very much more up to date than me as my research was all 20+ years ago and things have changed dramatically in that time. Interesting stuff none the less but, it moves on rapidly in leaps that said though stuff I was seeing and doing back then much of it seem's still not to have hit the run of the mill user.

Interpretation is one of the big problems I had with data mining output, not my problem a problem with users. I could send one of my bot's into a data warehouse to go and find regularly occuring, strange and unusual data groupings and connections but, what to make of them. My experience of watching users data mining in the big bad world is that the user set their query and then receive an answer. This is not data mining in the true sense and people don't seem to get this, it need's a whole diffent kind of person to look at the data mining results to intepret them and fathom what it means. About data mining results my Prof used to say and I quote, "when you see the results of DM you should be looking for the one that makes you say '**** me, I didn't know that' ".

Good luck with ethics. I taught a structured methods module to Doc's doing a masters in order to show them how to try and approach problems in a more structured engineering like way - basically to put a bit of logical order into their decision making approaches. The biggest issue they had with the use of expert systems and AI was ethical in how far should they hand off accountability and responsibility for decisions made by a box.
 
#27
With a very limited knowledge of computing, every time I hear or read about AI, I just don't believe in it. To me it seems that if you interact with a computer and it gives you information, it's a souped-up version of Conditional Formatting in Excel - you input something and the computer is programmed to give you a response against what you've said or typed.

That isn't AI, it's simply clever computer programming. Am I right or have I got the wrong end of the stick?
It's a little bit more sophisticated than that.

Intelligence, Artificial or otherwise it is not.
 
#28
To an extent a chatbot reminds me of the old apple scam about style over substance. The boxes and appearance is ever more pretty and attractive as times go by, but the actual service delivery and application is not great and whatever savings you make in implementation, has to be balanced to how much you want to irritate your customers.

But, the real deal is all focused on facial recognition especially with regards gauging emotions and the one thing I never dare ask my managers are the legal side. If in future world, a mistake is made and computers always make mistakes. GDPR and the odds of a massive rise in lawsuits, seem inevitable .
It goes beyond that. The mum (milf), of one of my lads mates, works for a company up in the NY area. we were talking about the stuff one day and she showed me a couple of client demo's on her laptop, shall we say very discerning clients. Not only facial recognition, combine it with gait analysis, body language analysis, movement between and acknowledgement of other individuals in the crowd and also luggage analysis, ie backpack, briefcase, roll along, carrier bag. They reckon that not only can they can ID individuals better than just facial recognition, they can also ID threats, ie scrotes, thieves and a hint was made at suicide bombers.

A UK based company has for many years had a system which locks onto your cellphone when you enter a premises, couples that to facial recognition and other stuff. Their application is primarily directed at the retail sector but, it has other uses too......oh yeah, the scary bit, they are directly linked to the consumer credit checking db's, so they know all about you already.

For the tinfoil hat brigade: Wear baseball hats, sunglasses, scarves and pay in cash.
 
#29
It goes beyond that. The mum (milf), of one of my lads mates, works for a company up in the NY area. we were talking about the stuff one day and she showed me a couple of client demo's on her laptop, shall we say very discerning clients. Not only facial recognition, combine it with gait analysis, body language analysis, movement between and acknowledgement of other individuals in the crowd and also luggage analysis, ie backpack, briefcase, roll along, carrier bag. They reckon that not only can they can ID individuals better than just facial recognition, they can also ID threats, ie scrotes, thieves and a hint was made at suicide bombers.

A UK based company has for many years had a system which locks onto your cellphone when you enter a premises, couples that to facial recognition and other stuff. Their application is primarily directed at the retail sector but, it has other uses too......oh yeah, the scary bit, they are directly linked to the consumer credit checking db's, so they know all about you already.

For the tinfoil hat brigade: Wear baseball hats, sunglasses, scarves and pay in cash.
The success rate of facial rec has improved massively and it has clear uses for the intelligence community who don't have to justify themselves to ambulance chasing lawyers. But to steal a pun from anchorman sex panther: "They've done studies, you know. 60% of the time, it works every time."
 
#30
It goes beyond that. The mum (milf), of one of my lads mates, works for a company up in the NY area. we were talking about the stuff one day and she showed me a couple of client demo's on her laptop, shall we say very discerning clients. Not only facial recognition, combine it with gait analysis, body language analysis, movement between and acknowledgement of other individuals in the crowd and also luggage analysis, ie backpack, briefcase, roll along, carrier bag. They reckon that not only can they can ID individuals better than just facial recognition, they can also ID threats, ie scrotes, thieves and a hint was made at suicide bombers.

A UK based company has for many years had a system which locks onto your cellphone when you enter a premises, couples that to facial recognition and other stuff. Their application is primarily directed at the retail sector but, it has other uses too......oh yeah, the scary bit, they are directly linked to the consumer credit checking db's, so they know all about you already.

For the tinfoil hat brigade: Wear baseball hats, sunglasses, scarves and pay in cash.
Any pictures of the MILF, asking for a friend?
 
#31
Any pictures of the MILF, asking for a friend?
Early 40's, 5'4" ish, arrse length brunette hair, bit of a bubble butt, nice face, well do able.

Sadly no pic's.
 
#32
I have seen AI used to great effect in medicine: being much more accurate and reliable than the best doctors at identifying cancerous cells, and better predicting the side-effects of new medicines (as just two examples).
No you haven’t. You’ve only seen rigged demos. Same as self-driving cars.
 
#33
The success rate of facial rec has improved massively and it has clear uses for the intelligence community who don't have to justify themselves to ambulance chasing lawyers. But to steal a pun from anchorman sex panther: "They've done studies, you know. 60% of the time, it works every time."
When I was in academia the circles I used to move in online were forever offering free facial rec, gait analysis and fingerprint analysis AI software. There used to be some good basically academic websites where you could download top shelf freeware, a bit like Ubuntu and linux a fair few of those were AI bundles people had developed during their PhD's to help with their PhD or just for fun.

The chap to look up for facial rec used to be Igor Aleksander from Imperial College. He was front end on the research back in the 80's & 90's and I doubt if there is much out there not partly built on the back of stuff he pionered.
 
#36
If I've got to the stage where I need to get in touch with a company I want to talk to a real person who can sort my problem out.
But only after you've pressed buttons 1, 3, 2, #, 1, 4, 1 and # then listened to, "You are moving forwards in the queue" for half a life time.
 
#37
No you haven’t. You’ve only seen rigged demos. Same as self-driving cars.
I don't know if it was trolling or gen, but I saw someone ask about the basics of writing code for fly by wire for cars, and you could feel the face palming before anyone replied

I hope it was just someone trolling...
 
#39
Early 40's, 5'4" ish, arrse length brunette hair, bit of a bubble butt, nice face, well do able.

Sadly no pic's.
Your mission, should you choose to accept it...
 
#40
The interesting bit of AI is in the "I" bit. Say there's a chatbot, and it has one of three opening lines:

"Yo mug, we've made $50K out of idiots like you already today. How can we make it $51K?"

"I see you're 19. We've helped 39 people in your age group save money on car insurance already today. What can we do to make it 40?"

"Thanks for contacting us, how can we be of help?"

If the algorithm then works out the response rates and adjusts the opening gambit accordingly, it's doing its job. Better yet would be a report detailing the questions and response rates, suggesting more refinement on the most successful question, for each demographic.

I know this is a ridiculously simple example, but machine learning is where it's at. The machine needs to make its own decisions to be "AI".

"I see you are female, weigh 250lbs, and are between the ages of 14 and 50. Are you pregnant or just fat?" No-one need get hurt :)
 

Similar threads


Top