Serving the Quantitative Finance Community
-
- katastrofa
- Posts: 7949
- Joined:
- Location: Event Horizon
Re: If you are bored with Deep Networks
I don't know what the rest of the world is doing, but recently I've been busy teaching a very small and simple pre-trained LLM to explain to the user why my model predictions are as they are. It's quite valuable in two ways - to build trust and provide transparency, but also to address the regulatory expectation of interpretability and explainability of the AI model outputs.
-
- Cuchulainn
- Posts: 23029
- Joined:
Re: If you are bored with Deep Networks
Am I missing something? This software has been around since the 1980s
(BTW bad customer experience with KPN).
Once the due diligence was complete, it was time for the KPN team to start building the new environment and experiment with it.
Then, in January 2024, the first use case finally went live.
“Our first ever solution focuses on customer services and aims to make life easier for the team working there,” says Jörg Kramer. “This department handles thousands of calls a day, and its only visual record of them is a transcript that’s produced at their end automatically.
“To help cut down time reading through the transcript, we’ve developed a technology that creates detailed summaries of transcripts right after the call.”
Each summary produces four different types of information: why the call took place, what the agent did to support the customer, the outcome of the call and, if applicable, any scheduled follow-ups.
“That is revolutionary for us,” he says. “When you run a customer services operation, making the customer feel like you have their undevoted attention is crucial. GenAI allows us to do that and more – helping us to empower our agents to deliver the best possible services and make our customers happy.”
As next step for this solution, continues Bardelloni, GPT will generate additional takeaways tailored to the call center desk handling the call. We are also allowing our analysts to define and experiment with new takeaways for their desks,” he adds.
“At the same time, we are setting up an automated pipeline which evaluates the quality of the generated summary - ultimately allowing us to intervene very quickly in case of performance deterioration.”
(BTW bad customer experience with KPN).
Once the due diligence was complete, it was time for the KPN team to start building the new environment and experiment with it.
Then, in January 2024, the first use case finally went live.
“Our first ever solution focuses on customer services and aims to make life easier for the team working there,” says Jörg Kramer. “This department handles thousands of calls a day, and its only visual record of them is a transcript that’s produced at their end automatically.
“To help cut down time reading through the transcript, we’ve developed a technology that creates detailed summaries of transcripts right after the call.”
Each summary produces four different types of information: why the call took place, what the agent did to support the customer, the outcome of the call and, if applicable, any scheduled follow-ups.
“That is revolutionary for us,” he says. “When you run a customer services operation, making the customer feel like you have their undevoted attention is crucial. GenAI allows us to do that and more – helping us to empower our agents to deliver the best possible services and make our customers happy.”
As next step for this solution, continues Bardelloni, GPT will generate additional takeaways tailored to the call center desk handling the call. We are also allowing our analysts to define and experiment with new takeaways for their desks,” he adds.
“At the same time, we are setting up an automated pipeline which evaluates the quality of the generated summary - ultimately allowing us to intervene very quickly in case of performance deterioration.”
Re: If you are bored with Deep Networks
I know I'm asking the same question over and over again, but have you ever managed to get *useful* answers to your questions using AI bots on a websites e.g. how to terminate your electricity contract, how to get your phone number off directory, or any such not advanced questions?
-
- Cuchulainn
- Posts: 23029
- Joined:
-
- Cuchulainn
- Posts: 23029
- Joined:
-
- katastrofa
- Posts: 7949
- Joined:
- Location: Event Horizon
Re: If you are bored with Deep Networks
I just skimmed it. It looks like a really nice project, but suitable mostly for static multivariate data. I myself work mostly with complex timeseries, where some more specialized methods (and extra carving) are necessary. I will try it out! Thanks!
-
- Cuchulainn
- Posts: 23029
- Joined:
Re: If you are bored with Deep Networks
PSEUDO AI - WHY PEOPLE BELIEVE IN FALSE PROPHETS LIKE KNOWN AI GURUS
"The world wants to be deceived, so let it be deceived." https://lnkd.in/eE_QiK6g
INTRODUCTION
"The pontifex maximus Scævola thought it expedient that the people should be deceived in religion; and the learned Varro said plainly, that there are many truths, which it is useless for the vulgar to know; and many falsities which it is fit the people should not suppose are falsities." https://lnkd.in/e5dsEMmy
Only this fame is fleeting, because for better or worse, all these charlatans will be exposed sooner rather than later, even if most people acknowledge it, they "still believe in them because they wanted to."
PSEUDO AI
The most likely explanation should be the truth, according to the new AI gurus. So, statistics combined with empiric algorithms have to be the truth. Even this is just an assumption, most people believe it. Why? Because the self-entitled AI gurus of Meta, Google, etc., are claiming the secret of intelligence, which is exactly what others want to hear.
Reading tea leaves has always attract the masses, even in the Roman Empire there were those who read the truth from bones, aka predict the future, same as the pseudo AI claims today. INFERENCE - DRAWING CONCLUSIONS, BASED ON WHAT? https://lnkd.in/eWTqzY6p
SIMPLE EXPLANATIONS FOR A COMPLEX WORLD
The world is complex, and our evolutionary savannah brain is not designed to grasp all the situations we encounter in life. Human technology has changed the entire landscape, and massive new knowledge via the Internet has increased exponentially.
All of this is too much for our brains to handle, so we leave understanding aside and try to find "simple solutions to complex problems." Why is that? The reason is simple: we don't want to deal with "details", we just want to deal with the conclusion. For this reason, scammers have a high appeal because they offer simple explanations that read, "IF this situation occurs, THEN this action will suffice."
DECEPTION
"Deception or falsehood is an act or statement that misleads, hides the truth, or promotes a belief, concept, or idea that is not true. It is often done for personal gain or advantage.
Deception can involve dissimulation, propaganda and sleight of hand as well as distraction, camouflage or concealment. There is also self-deception, as in bad faith. It can also be called, with varying subjective implications, beguilement, deceit, bluff, mystification, ruse, or subterfuge." https://lnkd.in/eFEdPqwT
EPILOGUE
All these "AI believers" will be deceived one day, and then reality will replace the nostalgia of simplicity. They have not accepted the truth and therefore have not adapted to the new conditions, namely that cognitive AI will be a major challenge for most of humanity.
As a result, they will lose their jobs and mass unemployment will be the new norm. Politicians will also do nothing, because they too were believers.
Mundus vult decipi, ergo decipiatur - Wikipedia
"Die weltt die will betrogen syn"
"The world wants to be deceived, so let it be deceived." https://lnkd.in/eE_QiK6g
INTRODUCTION
"The pontifex maximus Scævola thought it expedient that the people should be deceived in religion; and the learned Varro said plainly, that there are many truths, which it is useless for the vulgar to know; and many falsities which it is fit the people should not suppose are falsities." https://lnkd.in/e5dsEMmy
Only this fame is fleeting, because for better or worse, all these charlatans will be exposed sooner rather than later, even if most people acknowledge it, they "still believe in them because they wanted to."
PSEUDO AI
The most likely explanation should be the truth, according to the new AI gurus. So, statistics combined with empiric algorithms have to be the truth. Even this is just an assumption, most people believe it. Why? Because the self-entitled AI gurus of Meta, Google, etc., are claiming the secret of intelligence, which is exactly what others want to hear.
Reading tea leaves has always attract the masses, even in the Roman Empire there were those who read the truth from bones, aka predict the future, same as the pseudo AI claims today. INFERENCE - DRAWING CONCLUSIONS, BASED ON WHAT? https://lnkd.in/eWTqzY6p
SIMPLE EXPLANATIONS FOR A COMPLEX WORLD
The world is complex, and our evolutionary savannah brain is not designed to grasp all the situations we encounter in life. Human technology has changed the entire landscape, and massive new knowledge via the Internet has increased exponentially.
All of this is too much for our brains to handle, so we leave understanding aside and try to find "simple solutions to complex problems." Why is that? The reason is simple: we don't want to deal with "details", we just want to deal with the conclusion. For this reason, scammers have a high appeal because they offer simple explanations that read, "IF this situation occurs, THEN this action will suffice."
DECEPTION
"Deception or falsehood is an act or statement that misleads, hides the truth, or promotes a belief, concept, or idea that is not true. It is often done for personal gain or advantage.
Deception can involve dissimulation, propaganda and sleight of hand as well as distraction, camouflage or concealment. There is also self-deception, as in bad faith. It can also be called, with varying subjective implications, beguilement, deceit, bluff, mystification, ruse, or subterfuge." https://lnkd.in/eFEdPqwT
EPILOGUE
All these "AI believers" will be deceived one day, and then reality will replace the nostalgia of simplicity. They have not accepted the truth and therefore have not adapted to the new conditions, namely that cognitive AI will be a major challenge for most of humanity.
As a result, they will lose their jobs and mass unemployment will be the new norm. Politicians will also do nothing, because they too were believers.
Mundus vult decipi, ergo decipiatur - Wikipedia
"Die weltt die will betrogen syn"
-
- Cuchulainn
- Posts: 23029
- Joined:
Re: If you are bored with Deep Networks
Inferring something from data alone is not intelligence, it's a sales pitch. The cognitive element is missing.
Re: If you are bored with Deep Networks
The AI gurus started in 350BC when Aristotle was talking about "Eikos", the likeness of truth. The Greeks were quite happy with probability without numbers. Non of it has changed that much since.
I'll spare you my entire talk.....
I'll spare you my entire talk.....
Website: https://jasonbelldata.com
Linkedin: https://www.linkedin.com/in/jasonbelldata/
Author of Machine Learning: Hands on for Developers and Technical Professionals (Wiley).
Contributor: Machine Learning in the City (Wiley).
Linkedin: https://www.linkedin.com/in/jasonbelldata/
Author of Machine Learning: Hands on for Developers and Technical Professionals (Wiley).
Contributor: Machine Learning in the City (Wiley).
-
- Cuchulainn
- Posts: 23029
- Joined:
Re: If you are bored with Deep Networks
The AI gurus started in 350BC when Aristotle was talking about "Eikos", the likeness of truth. The Greeks were quite happy with probability without numbers. Non of it has changed that much since.
I'll spare you my entire talk.....
-
- Cuchulainn
- Posts: 23029
- Joined:
Re: If you are bored with Deep Networks
AI-assisted targeting in the Gaza Strip
https://en.wikipedia.org/wiki/AI-assist ... Gaza_Strip
Khlaaf went on to point out that such a system's decisions depend entirely on the data it's trained on, and are not based on reasoning, factual evidence or causation, but solely on statistical probability.
https://en.wikipedia.org/wiki/AI-assist ... Gaza_Strip
Khlaaf went on to point out that such a system's decisions depend entirely on the data it's trained on, and are not based on reasoning, factual evidence or causation, but solely on statistical probability.
Re: If you are bored with Deep Networks
Sadly such things have been on other borders for a long time.AI-assisted targeting in the Gaza Strip
Website: https://jasonbelldata.com
Linkedin: https://www.linkedin.com/in/jasonbelldata/
Author of Machine Learning: Hands on for Developers and Technical Professionals (Wiley).
Contributor: Machine Learning in the City (Wiley).
Linkedin: https://www.linkedin.com/in/jasonbelldata/
Author of Machine Learning: Hands on for Developers and Technical Professionals (Wiley).
Contributor: Machine Learning in the City (Wiley).
-
- Cuchulainn
- Posts: 23029
- Joined:
Re: If you are bored with Deep Networks
TrueSadly such things have been on other borders for a long time.AI-assisted targeting in the Gaza Strip
-
- Cuchulainn
- Posts: 23029
- Joined:
Re: If you are bored with Deep Networks
“I'd also like to acknowledge my students (…) they've gone on to do many great things. I'm particularly proud of the fact that one of my students [Ilya Sutskever] fired Sam Altman.“
The Godfather of Deep Learning who later became the Godfather of AI also added:
"Over time it turned out that Sam Altman was much less concerned with safety than with profits and I think that's unfortunate" - Geoffrey Hinton
This has to be the most brutal yet funniest thing ever said in AI.
The Godfather of Deep Learning who later became the Godfather of AI also added:
"Over time it turned out that Sam Altman was much less concerned with safety than with profits and I think that's unfortunate" - Geoffrey Hinton
This has to be the most brutal yet funniest thing ever said in AI.