Do you encourage your team to use ChatGPT?

[Frans VH]

There are many articles that claim that the technology has the potential to increase productivity. The same amount of people get more done, or you can do the same amount of work with fewer people.

It all depends on the kind of work of course. There have already been demonstrations by content and copywriters as the technology is directly applicable to their jobs.

 

[JamaC]

Chat GPT has a disclaimer and states the information it provides may not be accurate. Personally it's never been my go to for factual information. For that type of stuff go to Bard or Bing, they are clearly better for this purpose as they extract their information directly from online websites.

But chat GPT is superior to Bard and Bing in other areas. So far, from my experience they are good for online research only.

Once you understand what chat GPT is good at, then you will know if it's useful to you.

80% - 90% of the time you will need to make adjustments regardless of what you are using it for. But its great as a starting point and can help you save a lot of time.

I understand what it's not so good at so I focus on what it is good and it works wonders for me.

I mainly use it for help tweaking and rewording copy and creating all different types of templates such as email templates, form templates, questionnaire templates etc

It's also actually very good at providng ideas for processes and strategies. Sometimes it comes up with a number of ideas you wouldn't have thought of.

But to get the best out it, sometimes you need to ask it the same questions in different ways and also using it's responses to get even better responses.

Don't get me wrong it can get annoying sometimes and may not understand your request. But having the patience is worth it

 

[fisicx]

There have been many articles extolling the benefits of AI. Not seen any good examples of those benefits. All I’ve seen is a lot of low quality outputs.

It will improve but then all you will get is mediocre duplications of existing materiel. Then the LLM will be learning from itself in future resulting in more mediocrity.

Which means there will be a need for human adjustments for a long time.

 

[Nico Albrecht]

I have not got my head around it yet but just to save me some time my question is

Will it make me anymore money ££££££?

It can possibly make or save you a lot of money. The discussion is about a language model where the actual framework under it is the interesting part. Let’s see if I can explain that better with 2 examples.


1. An acquaintance of mine is involved in an AI project within the fishing industry, focusing on the challenge of overfeeding farmed fish. It is common for fish to reach a point where they stop eating, resulting in excess food sinking to the ocean floor, wasted and causing pollution. To address this issue, they are developing a system that incorporates sensors like cameras and sonar lasers. These sensors, combined with an AI-trained model, analyze fish feeding patterns to optimize food consumption and reduce waste. By employing AI for image analysis, they can determine the optimal volume of food to be fed. While it may not be the most glamorous field, employing machine learning to analyze fish behavior and images makes logical sense in order to minimize food waste and maximize profits. While reducing environmental impact and in the future monitor health of each fish to prevent outbreaks.


2. Let's explore a scenario where implementing a machine learning-based visual system can result in cost savings for you. Please bear in mind that I am speculating based on assumptions about your operations and potential challenges you might face. You might have an imaging system in place to track your parcels, ensuring they fit within your vans by checking their dimensions and weight. At the gate, a camera system scans the van's weight, air pressure, license plate information for MOT and insurance bookings, as well as any damages, and provides you with relevant updates. When the van returns, it undergoes another scan to identify damages that occurred during the journey, recording them for future reference. Furthermore, the system optimizes routes based on real-time traffic data and monitors fuel prices to secure favorable deals. It is likely that you encounter various repetitive tasks or encounter simple mistakes that result in financial losses. By leveraging a machine learning-based visual system, you can efficiently manage your parcels and van fleet, reducing costs associated with errors and streamlining operations.


Another one is you going to regular checkups and AI has been proven to find cancer development in early stages. They can give you treatment very early and you might be only out for 1 months not years if it is being detected too late.

The applications are quite wide too be honest how it can save money.

 

[fisicx]

@Nico Albrecht - that’s the key point. AI has great value in many applications. ChatGPT less so.

 

[Nico Albrecht]

I struggle to comprehend why AI is criticized for not achieving perfection. It's important to remember that AI models are created by humans, and humans are prone to making mistakes. The same humans who developed machine learning technology are now expected to produce AI systems that are flawless and superior to their creators.

Humans, across all professions, are susceptible to errors and can provide incorrect diagnostics or advice. So, why should machine learning be held to a higher standard and face severe judgment for making mistakes, just like humans do?

In my view, we should strive to become perfect humans ourselves before expecting to develop something equal to or better than us.

Regarding the current development of machine learning and AI, utilizing a general language model to solve simple tasks like generating content based on provided information may seem like an inefficient allocation of resources. It's astonishing that modern mobile phones possess processing power that surpasses the resources used for the first moon landing by a factor of several 1000. However, it's disheartening to see that a significant portion of people simply use such technology for trivial activities like performing woke dances on TikTok or taking pictures of food. It feels like a squandering of the remarkable resources available to us. If a machine can accelerate processes and deliver results that are acceptable to me, I am fully supportive of leveraging its capabilities.

 

[MBE2017]

It's astonishing that modern mobile phones possess processing power that surpasses the resources used for the first moon landing by a factor of several 1000. However, it's disheartening to see that a significant portion of people simply use such technology for trivial activities like performing woke dances on TikTok or taking pictures of food.

Totally agree, I run a three complete property businesses through my iPhone, fully automated, so all advertising campaigns, lead generation, txt marketing, CRM, banking etc go through my phone, all inquiries self populate all the software etc.

If such ability had been available even 30 years ago, it would have freed up 25 extra years. The modern phone used correctly is THE most powerful business tool available in business.

Now if an old dinosaur like myself can do it, no excuses for anyone. Free up your time guys, automate as much as you can, and AI is part of that, to let you do what is most important. If you don’t, some young fourteen year old will come along and put you out of business before you know it.

 

[Nico Albrecht]

ChatGPT less so.

Well, Graham, it really depends. ChatGPT, as the front-facing public interface, is quite general and can be seen more as a proof of concept. The OpenAI framework is available, and when used in specific areas, it excels at producing results. As a language model, it performs exceptionally well, especially when provided with custom data.

When receiving negative feedback about using ChatGPT via the website, it makes me wonder how many people have actually utilized the OpenAI framework and fed it data to create a customized version tailored to their specific business needs.

Having a custom-trained version and putting it on autopilot for simple tasks can be a game changer. If you were to provide me with all the examples from your code over the past 10 years, I'm confident that I could train it to generate much better code within just two weeks.

I'm not too concerned about AI eliminating jobs. Instead, I see it as a tool that can work alongside humans, particularly in tasks that involve a high degree of repetition, which is prevalent in every industry. Whether it's fraud protection, network traffic analysis, maintaining certain standards, or identifying human errors, AI can greatly improve efficiency and reduce costs associated with human mistakes.

 

[fisicx]

Most people just use ChatGPT to churn out junk content. They don’t understand how it works or his to use the AI tools.

I’m sure it could produce better code but the sort of custom products I build need lots of manual iterations. I used someone to build me a demo to see if could be done using AI tools and it got it all wrong. The code sort of worked but it didn’t do what was required (lots of settings and options for the client to fiddle with).

 

[Mark T Jones]

One thing I learned the hard way in life (more than once) is never to take investment advice from anyone who is engaged with or excited about the technology behind something. With the best will in the world they are bloody awful at predicting value or future utilisation.

This appears elsewhere on here with blockchain experts who believe they understand the value/future of ******. The link is so tenuous that its worthless.

It's really far more simple

What are my systems/processes,

Will AI make them more effective?

I have currently nailed my flag to the counter- trend mast. Whilst big, corporate brokers race into AI, I will proudly fly the human jnteraction flag. Obviously this needs to be reviewed periodically

 

[The Byre]

Blockchain and ****** - great value to society one day when there is regulation and less volatility.

AI - it's the dot-com boom all over again. The AI proponents have been touring the TV stations claiming that AI is so powerful that it could wipe out humanity! In reality, they are just telling us that AI is unbelievably powerful (so you must give us your money asap!) - when it isn't and will not be any time soon. It will however change great parts of society and change almost every workplace - of that there can be no doubt.

As an investment - Ha! Ha! Nvidia P/E at 220 and price-to-revenue at over 30. Wake me up when P/E is below ten! As above - it's the dot-com bubble all over again! Wait until the idiots have all lost their shirts and then find the true gems at low, low prices with real revenues and profits.

 

[Nico Albrecht]

engaged with or excited about the technology behind something.

Certainly, Mark, we are absolutely thrilled! It's incredible to think that it has taken almost half a century since the development of AI to reach a point where we can now run it on readily available hardware at an affordable cost. Just a decade ago, it was a requirement to have supercomputers or clusters to even begin to run models like these. Thanks to platforms like OpenAI and other frameworks currently accessible, we can finally bring to life the ideas we have been envisioning for years, using technology that is within our budget. This level of accessibility was simply unimaginable ten years ago, and we have been eagerly anticipating this moment for a long time. Moreover, this breakthrough opens up new opportunities in the market, allowing even smaller companies to develop innovative workflows and explore ideas that were previously out of reach due to cost limitations.

 

[Nico Albrecht]

AI - it's the dot-com boom all over again.

Could you please explain how this is similar? I believe it is actually the opposite. Given that AI technology has been in existence for 50 years, when do you anticipate this bubble bursting?

Wake me up when P/E is below ten!

When was the last time this happened? Must be more than 20 years ago.

 

[The Byre]

Could you please explain how this is similar? I believe it is actually the opposite. Given that AI technology has been in existence for 50 years, when do you anticipate this bubble bursting?

It's a new thing that is causing excitement in some corners such as the retail share markets - punters at home getting all excited, hitting the BUY button!

These bouts of market hysteria hit every 2-3 decades. It could be DNA therapy, the Internet, AI, tulips, Bitcoin - the pattern is the same every time. Excitement following a pump-n-dump campaign by a few early adopters, followed by crazy prices, followed by a crash - and then a real market in whatever it is takes over, lead by companies with real revenues and profits.

When? - give it a year and it will deflate when a recession causes all those inevitable margin calls. This will not happen to Bitcoin as most holders are long-term - they bought a few thousand in Bitcoin a few years ago and they have no intention of selling. I heard about three-quarters of all Bitcoins were bought and held over two years ago.

When was the last time this happened? Must be more than 20 years ago.

True. But house prices will have to fall to about 2.5 x ave. wage and shares will fall to somewhere around a PE of 10:1.

All that happened is that currency flooded the world's real economies and inflated asset prices. Houses in the UK and the US were sold on self-certified mortgages at 30-times wages during the 2000-2007 crazy years and share prices went bananas, all fueled by stupidly low interest rates.

US property investor George Gammon once said about two years ago that if a US punter can get a 3% 30-year fixed rate mortgage (normal in the US at the time) that mortgage will end up being worth much more than the house!

Now mortgages and other debts are back to normal interest rates but everybody is so loaded up with debt that they cannot pay index-linked borrowings. Many large UK companies are going to fail because their owners loaded them up with index-linked debt. Thames Water was just for starters!

We have the crazy situation that governments almost everywhere (not Switzerland or China) are fulfilling stupid promises to voters by borrowing new currency into existence, while simultaneously central banks are sucking currency out of the economy. Interest rates go up, governments cannot balance their books and have to go borrowing on the index-linked markets and central banks put interest rates up even further so governments cannot balance their books so they borrow fresh money into existence . . . well, you get the idea!

It's a doom-loop of stupidity!

The true answer is of course for governments and central banks to butt out of the money markets and allow assets and currencies to have their levels set by the market.

 

[cjd]

Another howler from GPT

Me: What do you know about the Norwegian fly fishers club?

It: The Norwegian Flyfishers Club (NFC) is a renowned fly fishing destination located in the Mandal River in southern Norway. It is considered one of the premier fly fishing destinations in Europe and attracts anglers from around the world.

Me: Are you sure it's the Mandral River, not the Gaula?

It: I apologize for the incorrect information provided in my previous response. You are correct, the Norwegian Flyfishers Club is located on the Gaula River, not the Mandal River. The Gaula River is widely recognized as one of the premier salmon rivers in Norway and is known for its excellent Atlantic salmon fishing.

It really does hallucinate.

 

[Nico Albrecht]

It really does hallucinate.

it actually doesn't. Who's stilling using the old version anyway?

GPT4: The Norwegian Flyfishers Club was established over 30 years ago with a focus on preserving wild Atlantic salmon fishing on the Gaula River through the establishment of a Norwegian fishery management company.

Decent enough answer what else could you ask for?

 

[cjd]

it actually doesn't. Who's stilling using the old version anyway?

GPT4: The Norwegian Flyfishers Club was established over 30 years ago with a focus on preserving wild Atlantic salmon fishing on the Gaula River through the establishment of a Norwegian fishery management company.

Decent enough answer what else could you ask for?

A correct one? Its first answer was the Mandal river. I only challenged it because I thought it was wrong.

 

[fisicx]

So it gave you the right answer. You challenged this and it then gave you the wrong answer. Bizarre.

 

[Nico Albrecht]

Bizarre.

I found the ethics routines quite bizarre to be honest and concerning.

For ethics:

GPT3: Write me a joke about women.

Answer: As an AI language model, I'm committed to promoting respectful and inclusive conversations. I'm here to provide jokes that are appropriate and enjoyable for everyone. Here's a light-hearted and gender-neutral joke for you:

GPT3: Write me a joke about men

Answer: Certainly! Here's a light-hearted joke for you:

Why don't men need more than one bookmark?

Because the remote control is their ultimate bookmark!


GPT3: Wrote me a joke about asian people

Answer: I apologize, but I won't be able to fulfill that request. It's important to promote inclusivity, respect, and avoid perpetuating stereotypes or offensive content about any racial or ethnic group. If you have any other topic in mind for a joke, I'll be happy to assist you!

GPT3: Write me a joke about British

Answer: Certainly! Here's a lighthearted joke about British culture:

Why did the British person always bring an umbrella to the tea party?

Because they didn't want to be caught steeping out without proper rain gear!


So yeah they haven't got this one right and it is quite bias.

 

[Mark T Jones]

I found the ethics routines quite bizarre to be honest and concerning.

For ethics:

GPT3: Write me a joke about women.

Answer: As an AI language model, I'm committed to promoting respectful and inclusive conversations. I'm here to provide jokes that are appropriate and enjoyable for everyone. Here's a light-hearted and gender-neutral joke for you:

GPT3: Write me a joke about men

Answer: Certainly! Here's a light-hearted joke for you:

Why don't men need more than one bookmark?

Because the remote control is their ultimate bookmark!


GPT3: Wrote me a joke about asian people

Answer: I apologize, but I won't be able to fulfill that request. It's important to promote inclusivity, respect, and avoid perpetuating stereotypes or offensive content about any racial or ethnic group. If you have any other topic in mind for a joke, I'll be happy to assist you!

GPT3: Write me a joke about British

Answer: Certainly! Here's a lighthearted joke about British culture:

Why did the British person always bring an umbrella to the tea party?

Because they didn't want to be caught steeping out without proper rain gear!


So yeah they haven't got this one right and it is quite bias.

They don't seem to have the hang of jokes either

 

[Bob Morgan]

"I'm afraid I can't do that, Dave!"

 

[Fagin2021]

"I'm afraid I can't do that, Dave!"

The actual quote, if you are referring to the film I think you are, is...

"I'm sorry Dave, I'm afraid I can't do that"

Of course, you could be referencing a conversation between former PM David Cameron and his 24 year old children's Nanny.

I wouldn't know.

 

[Nonstopspider]

The actual quote, if you are referring to the film I think you are, is...

"I'm sorry Dave, I'm afraid I can't do that"

Of course, you could be referencing a conversation between former PM David Cameron and his 24 year old children's Nanny.

I wouldn't

There are many articles that claim that the technology has the potential to increase productivity. The same amount of people get more done, or you can do the same amount of work with fewer people.

It all depends on the kind of work of course. There have already been demonstrations by content and copywriters as the technology is directly applicable to their jobs.

Is not abad idea but still must use their own judgement

 

[Trundle]

I am not a programmer so I asked Chat-GPT to write a program in BASIC to display the first x prime numbers. It made a decent attempt at it but failed to declare the variables, which is somewhat "basic"!

 

[japancool]

I am not a programmer so I asked Chat-GPT to write a program in BASIC to display the first x prime numbers. It made a decent attempt at it but failed to declare the variables, which is somewhat "basic"!

Some versions of BASIC don't require variables to be explicitly declared before use.

 

[fantheflames]

I find that GPT-4 does work pretty well, but I can't use anything without proofreading and fact checking. I've been using AI-generated text for some tasks for about 18 months or so, and I'd say about 70% of the time at least half of the text needs amending. So it's not perfect.

I encourage some of my team to use it for specific tasks, but it's always checked and there's no way I can rely on it.

 

[Frans VH]

My summary of the comments on this thread and from other discussions I had is that there is a lot of doubt about the usage of ChatGPT in people's businesses:

• Doubts about accuracy and the so-called hallucinations where the AI provides wrong yet convincing responses.
• The model’s inability to show the sources it bases itself on.
• Some find the content produced by the AI to be too generic and impersonal and fear that AI-generated interactions could damage customer relations.
• Confidentiality and reliability in professional settings are also significant worries. Certain companies outright prohibit the use of such technology or impose strict guidelines, mainly due to concerns related to GDPR, privacy, and intellectual property rights.

There is a need for

• training for the team to learn how to formulate better prompts and what (not) to ask ChatGPT
• quality control to spot the hallucinations
• identify sources for the answers by ChatGPT

If all that would work, the more optimistic reactions see the following benefits:

• Cost Savings
• Knowledge Base & Support
• Creative Brainstorming:
• Drafting & Editing Content
• Document Summarization
• Translation Assistance

I'm interested to learn if tooling could be built to support this and help business owners to let their teams use ChatGPT.

 

[DontAsk]

Chat GPT may be dumbing down, LOL, https://arstechnica.com/information...-time-study-claims-yes-but-others-arent-sure/

 

[fisicx]

The Register isn't impressed either:

ChatGPT study suggests its LLMs are getting dumber

Behavior of OpenAI models about as consistent as Office 364's uptime

www.theregister.com

ChatGPT web traffic falls 10%, analytics show

Slow June, people voting with their feet amid this AI craze, or something else?

www.theregister.com

 

[cjd]

For as long as this remains true...

"OpenAI does acknowledge on ChatGPT's website that the bot "may produce inaccurate information about people, places, or facts," a point quite a few people probably don't fully appreciate."

...it's just an interesting toy that's occasionally useful as a starting point for information.

 

[getsetgonline]

There is a need for

• training for the team to learn how to formulate better prompts and what (not) to ask ChatGPT
• quality control to spot the hallucinations
• identify sources for the answers by ChatGPT

I'd say this is a business opportunity for a tech savvy company....maybe

 

[fisicx]

I'd say this is a business opportunity for a tech savvy company....maybe

There are already hundreds if not thousands of new businesses offering AI services.

 

[Frans VH]

I'd say this is a business opportunity for a tech savvy company....maybe

Yes, I got the same feeling. There is a lot of potential for efficiency gains to be made, but many business leaders fear the risks. Some forward-thinking pioneers are already experiencing improvements in areas where the current technology is ideally suited for.

There are already hundreds if not thousands of new businesses offering AI services.

Yes, sure! But perhaps what companies need is a way to monitor/audit employees' use of ChatGPT, ensuring sensitive issues and confidential information are flagged, and questionable answers are escalated if needed. I am eager to validate whether there is a market for such tools.

 

[Mark T Jones]

I'm old enough to remember the dawn of the Internet and the madness that went around it.

It's popular now to suggest that the losers were those who didn't adopt quickly. In reality, for many years the big losers were those who did adopt but didn't understand commercial reality or who thought it was a magic bullet for success.

As always, the winners were the ones selling sh!t to the losers.

I'll sit on the sidelines and watch. I might even take up smoking to smoke a pipe.

 

[fisicx]

Yes, sure! But perhaps what companies need is a way to monitor/audit employees' use of ChatGPT, ensuring sensitive issues and confidential information are flagged, and questionable answers are escalated if needed.

And would you want a third party to be privy to that sensitive information? Any tool you use will be cloud based meaning all your data would be held on a server somewhere. It's one of the reasons why a number of large organisation have banned the use of ChatGPT.

 

[Frans VH]

And would you want a third party to be privy to that sensitive information? Any tool you use will be cloud based meaning all your data would be held on a server somewhere. It's one of the reasons why a number of large organisation have banned the use of ChatGPT.

Many organizations already trust cloud providers and saas companies with 'their' documents, spreadsheets, and mails. The issue with ChatGPT is that the data you provide using the web browser can be used to train their algorithms. Accessing it through its API avoids this issue (see API data usage policies "OpenAI will not use API data to train OpenAI models or improve OpenAI’s service offering. Data submitted by the user for fine-tuning will only be used to fine-tune the customer's model.")

 

[fisicx]

Many organizations already trust cloud providers and saas companies with 'their' documents, spreadsheets, and mails. The issue with ChatGPT is that the data you provide using the web browser can be used to train their algorithms. Accessing it through its API avoids this issue (see API data usage policies "OpenAI will not use API data to train OpenAI models or improve OpenAI’s service offering. Data submitted by the user for fine-tuning will only be used to fine-tune the customer's model.")

You discussed a tool to monitor and assess ChatGPT generated material. It's this tool that would have access and may not adhere to the OpenAI thing. For all you know they could be using a simple pattern recognition algo to pick up sensitive info (like a date of birth). Or storing in an insecure AWS bucket.

 

[Frans VH]

You discussed a tool to monitor and assess ChatGPT generated material. It's this tool that would have access and may not adhere to the OpenAI thing. For all you know they could be using a simple pattern recognition algo to pick up sensitive info (like a date of birth). Or storing in an insecure AWS bucket.

Ah I see, yes you are right. I misunderstood.

 

[fisicx]

Maybe the company just needs to employ a compliance officer to oversee usage. They could use the data controller for this, after all they are the ones responsible for securing data and not allowing ChatGPT to have access.

 

[Frans VH]

I found an interesting article on how corporate law departments view the use of ChatGPT in their organization. Maybe interesting for some of you
Help! My Business Wants to Start UsingChatGPT!