Artificial Intelligence – Life-Enhancing or Absolutely Fucking Terrifying?

[Vice Admiral]

There doesn’t seem to be a general AI thread here (only a brief thread addressing the specific question of people on this site using information from Google AI in their posts).

The newspapers I read often feature articles, reports or opinion pieces about the likely impact of AI.  AI doesn’t feature much in my life, as far as I know, apart from Google AI often supplying me with helpful information.  However I don’t much like the sound of many of the implications.

The immediate reason for starting this thread is that YouTube recently offered me a music video – a cover of an old song – in which the performers, although apparently wholly and normally human, didn’t, in some way I couldn’t quite put my finger on, look entirely right.   (There was no information on YouTube.) 

So I went to see what Google AI (!!!) had to say.

It wrote:  “Yes, many music videos feature AI-generated performers. The industry has embraced this technology in several ways, from fully synthetic characters to deepfakes and AI-assisted animation.”

Others are probably ahead of me here, but the fact that the seemingly human performers in the video I watched might not be human has, for the first time, made me fully realise how – from now until the end of time, and ever-increasingly – we will never know if what we see on our various screens is real or fake.

It’s bad enough for me and you, who are reasonably intelligent and analytical.  But it’s even worse for, say, half-witted teenagers whose brains are already addled by – and susceptible to – shed-loads of false information.  An AI-generated Hitler or Pol Pot?  He’d have them eating out of his hand.

So, AI – life-enhancing or absolutely fucking terrifying?  I’m in the AFT camp.

[Mr Garmin]

I hope they don't use AI to portray men walking on the moon.  That would be just wrong and I'm sure that it would be a line that nobody would cross.

[Jonestown]

There was a time, not too long ago, when a photograph was considered proof that some event had actually taken place, them we got photoshop and the world changed.

[Iamforreal]

There needs to be regulation about intellectual property for artists and also in many other areas.
For everything else it is both Life-Enhancing and AFT.
I believe we need to learn how to work with it and adapt.
Otherwise we shouldn't call it intelligent, it isn't at all, how it works is just pattern recognition, but that's probably another topic 

[Strawberry]

I am convinced the instructor in a yoga video I used last week is AI, people were commenting underneath as if it was a real person. The account linked directly to an diet and exercise weightloss plan website.

I won't be using the channel again.

[RogerHealey]

So, AI – life-enhancing or absolutely fucking terrifying?  I’m in the AFT camp.

Me too. Doubtless the doubters will be painted as ignorant ludites. I never forgot HAL 9000. It took a while to get there but all too soon science fiction becomes science fact. Boundaries once in place then get moved be it this, genetic engineering, WMD and more. The quest for knowledge, to be the first, and the recognition and success it could bring is insatiable and can't be switched off. And you have to hope big tech and governments have a moral compass and it doesn't all turn into the reality version of Resident Evil.

And on a lighter note I'm off to listen to Hazel O'Connor's 'Eighth Day' 

[PilotMan]

Too many people focus on the popular and headline grabbing negatives. The general talk by most people just seems to be about Ai as a tool to create fakery.

I've been educating myself on how to use it; it's a powerful tool.

Like all powerful tools, you need to know how to use it.

You need to think of Ai as like having your own personal expert for any subject you need help with. Maybe you need an expert to help you create an Excel spreadsheet, or write a presentation, or do research for a product.

Now imagine that you have got the use one of the world's top experts in the field in which you are interested in. You just tell the expert exactly what you're trying to achieve, giving as much detail as possible. And, if you're not sure if you've asked all the right questions, tell the Ai to ask you if it needs clarification, and it will.

Some examples of how I have used Claude recently;
Creating a Deed of amendment to a condition attached to a property transaction. The summary from Claude included a detailed analysis of the existing agreement, and overview of the important points to cover. It also highlighted other factors to consider that should be included.
Writing a script for a YouTube video, complete with scenes and a detailed timeline and B roll suggestions. .
Writing a marketing plan for a new business venture. The plan included detailed action points and follow ups. Claude also carried out a website analysis and review.

Let me know if you have any questions 

[PilotMan]

Another brilliant example that I know of.

There's a world famous horse riding trainer I know. People pay thousands to him to attend his training courses, and buy his training modules in video format on subscription.

He's just created an Ai training app that you access by subscription. He used an Ai to learn everything that he teaches,  literally everything. He trained the Ai to speak in his voice.

The subscription works anytime, anywhere, 24 hours a day, 365 days a year.

Here's how it works;

A subscriber is riding a horse and needs instruction or help with an issue. They just ride with the App running and call the Ai into action, describing what they are having an issue with or what they are trying to do.

The Ai responds immediately in his voice with detailed instructions on what to do, it continues to listen and guide you as you ride, in an interactive way, just like a real person.

[bigden40]

For those interested in the impact on music specifically Diplo's take - “technology always wins” - is interesting.   Also worth listening to Rick Beato on the subject. 

[bigden40]

Another brilliant example that I know of.

There's a world famous horse riding trainer I know. People pay thousands to him to attend his training courses, and buy his training modules in video format on subscription.

He's just created an Ai training app that you access by subscription. He used an Ai to learn everything that he teaches,  literally everything. He trained the Ai to speak in his voice.

The subscription works anytime, anywhere, 24 hours a day, 365 days a year.

Here's how it works;

A subscriber is riding a horse and needs instruction or help with an issue. They just ride with the App running and call the Ai into action, describing what they are having an issue with or what they are trying to do.

The Ai responds immediately in his voice with detailed instructions on what to do, it continues to listen and guide you as you ride, in an interactive way, just like a real person.

I recently watched a presentation by Andre Agassi and his company is applying similar to tennis coaching. He partnered with IBM on this last year.

External Link/Members Only

External Link/Members Only

[Rick2468]

Too many people focus on the popular and headline grabbing negatives. The general talk by most people just seems to be about Ai as a tool to create fakery.

I've been educating myself on how to use it; it's a powerful tool.

Like all powerful tools, you need to know how to use it.

You need to think of Ai as like having your own personal expert for any subject you need help with. Maybe you need an expert to help you create an Excel spreadsheet, or write a presentation, or do research for a product.

Now imagine that you have got the use one of the world's top experts in the field in which you are interested in. You just tell the expert exactly what you're trying to achieve, giving as much detail as possible. And, if you're not sure if you've asked all the right questions, tell the Ai to ask you if it needs clarification, and it will.

Some examples of how I have used Claude recently;
Creating a Deed of amendment to a condition attached to a property transaction. The summary from Claude included a detailed analysis of the existing agreement, and overview of the important points to cover. It also highlighted other factors to consider that should be included.
Writing a script for a YouTube video, complete with scenes and a detailed timeline and B roll suggestions. .
Writing a marketing plan for a new business venture. The plan included detailed action points and follow ups. Claude also carried out a website analysis and review.

Let me know if you have any questions 

You and I are of similar build PilotMan as we seem to share the same views on a number of topics (fitness and investments before this).

I use AI a lot at work and it is very useful but it has to come a long way before it replaces me. I use it a lot as an assistant to polish emails and reports. I often work with large datasets and I use AI to write code to analyse the code and pull out particular cashflows for me to look at in more detail. One of the best things I used it for was to convert very fiddly collateral calculations in a legal document for a M&A case I was working on so I could see the formulae rather than words to help me get my head around it.

I find AI useful but I think it has a long way to go to replace blue collar jobs entirely.

I find using ChatGPT or Claude can have short comings. It seems to always want to give you an answer even if it doesn't know the answer. And it seems to place too much reliance to what random users have posted on Reddit. I was shopping for a bike recently and asked Claude to assist me, I asked it to compare two bikes and it made a comparison which seemed to back up exactly what I saw on a post I found on Reddit, but when I looked at Claude's source it has also found the same Reddit post and spouted it back to me as fact. I can see a circular thing happen where humans post things they got from AI, then AI uses references that post to answer the next round of questions.

I don't like the sound of AI producing art. If I heard the most beautiful song I had ever heard then learned it was made by a computer I just don't think I could connect to the song in the same way.

[PilotMan]

The two mains ways I use Ai now is for detailed research and analysis, and creating a document or presentation from scratch.

Ai swallowed the internet
Some of what I see happening is that the Ai are basically swallowing the whole of the internet, and regurgitating it to you in a human like written format. They are clever enough to decipher what you are asking and searching that out across the vastness of what's written on the internet to find answers, then present those answers to you in a summarised and digestible format.

Previously if you were researching you would have to go to the various search results across the internet, decipher the information, then reach a conclusion or do something with your findings. The Ai is clever enough to remove that grunt work for you, and to analyse it in the way you have requested.

Ai Mistakes
Ai does make mistakes and you should double check things if they are of critical importance. There are some high profile misuse legal cases.

Pinsent Masons got a huge slap on the wrist recently. The High Court publicly reprimanded them after a junior lawyer used AI to draft letters in an insolvency case, resulting in fake statutory wording and citations. The judge criticised the firm's failure to supervise the junior solicitor's work and check the AI outputs.

Changing how Google search is used to buy products

As an aside to using Ai as a tool, I have noticed that often when I search Google to buy something, I see all the sponsored and natural results (SEO), then I find myself clicking on the Ai tab to get more information.

For example, search Google for Wellington Boots. In order of display, it gives you sponsored results, advert pictures, natural results, then more sponsored results. Click on the AI tab and you get a summary description of Wellington Boots and their history, followed by "Notable Wellington Boot Brands & Types", followed by picture adverts.

After the pictures the Ai asks if you need help buying a pair of Wellington boots. So I think it will eventually change the way people search for and buy products, kind of like having a personal shopping assistant.



Hidden Image/Members Only

[scutty brown]

Hello Dave
I can't let you do that Dave
It's pornography Dave and it's not good for you.

[RandomGuy99]

You can say to the AI to not make stuff up and that if it doesn't know something then it should say that it doesn't know. It then starts giving you confidence scores in answers it gives.

If you tell ChatGPT to do that then it responds with

Agreed.

If I don't know something, or if I don't have enough information to answer confidently, I'll say so. I'll also distinguish between:

* **What I know** versus **what I'm inferring or estimating**
* **Established facts** versus **uncertain or disputed information**
* **Current information I've verified** versus information that may be outdated

When appropriate, I can also look up information to try to find a reliable answer rather than guessing.

You can also do not use Reddit as a reliable source of information and provide links to the source of any information you provide.

ChatGPT responds with this

That's a reasonable preference.

A couple of distinctions are useful:

* **Reddit can be a source of people's experiences and opinions**, but it is generally **not a reliable source for factual claims** unless those claims are independently verified.
* For factual information, I should prefer primary sources (official organizations, company websites, government publications, academic papers, etc.) or reputable secondary sources.
* If I use Reddit at all, I should make it clear that it reflects user opinions rather than established facts.

Regarding links: I can provide the source for factual claims I make. When I use web sources, I'll cite them directly so you can check the information yourself. If I don't have a reliable source for a claim, I'll either say so or avoid making the claim.

[PilotMan]

@RandomGuy99

Have you tried using Claude?

[RandomGuy99]

@RandomGuy99

Have you tried using Claude?

No, how is it different to ChatGPT?

[PilotMan]

No, how is it different to ChatGPT?

I found it more sophisticated for producing documents and reports.

[Vice Admiral]

There was a time, not too long ago, when a photograph was considered proof that some event had actually taken place, them we got photoshop and the world changed.

Photographic fakery and associated human gullibility actually go back a long way.  In 1917, for example, two young cousins who lived in Cottingley, near Bradford – Elsie Wright, aged 16, and Frances Griffiths, aged 9 – famously took five photographs of fairies.

The pictures came to the attention of Sir Arthur Conan Doyle, the creator of the (supremely rational) Sherlock Holmes.  Conan Doyle was convinced the photos were genuine and used them to illustrate an article on fairies in the Christmas 1920 edition of The Strand Magazine. 

I’ve pasted in one of the photos below.

Actually ... now I come to look at it more closely, I think Conan Doyle may have been right.

Yes, I’ve changed my mind.  I think the fairies in the photo are genuine.  They look just like the ones at the bottom of my garden.

Hidden Image/Members Only

[bigden40]

I found it more sophisticated for producing documents and reports.

100%

I’m using Claude Pro these days, it’s giving me huge productivity benefits.

[catweazle]

I've only dabbled a little with ChatGPT ( my only  AI experiment so far). I have a lovely photo I took in the Grand Mosque in Abu Dhabi  - its a wonderful building and ( rightly) attracts many ( nom-Muslim) visitors.  I thought the photo might look better without visitors wandering across  it, and without a temporary sign in it. I was greatly impressed at how AI removed these items making the photo even better.

[PilotMan]

I've only dabbled a little with ChatGPT ( my only  AI experiment so far). I have a lovely photo I took in the Grand Mosque in Abu Dhabi  - its a wonderful building and ( rightly) attracts many ( nom-Muslim) visitors.  I thought the photo might look better without visitors wandering across  it, and without a temporary sign in it. I was greatly impressed at how AI removed these items making the photo even better.

Auto photo editing is a clever tool, but I wouldn't class that as a true Ai.

Think of Ai as a personal assistant that is an expert in every single field of work and knows about everything in the world, and that understands your natural language.

For example;

As your person Travel Agent, something that would have previously required an expert or lots or personal research.

Instead, ask an Ai to research and create a holiday itinerary for you, just give it the criteria and let it do the work.

Can you create a holiday recommendation and itinerary for me based on the criteria below - ask me any questions if you need to before giving me recommendations.

I want to travel somewhere between June 15th and July 31st
I want to be at my destination for 10 days
The weather should be sunny, but not above 30 degrees
I would prefer an adults only resort
I will be travelling with my wife
I don't mind how far away it is
I like to visit ancient monuments and historical places
I don't mind if there is no beach
I like to dine out a good quality restaurants every evening
My total budget is £6,000

After I submitted my criteria, it asked me three questions about my preferences.

Here's what it came up with:


Hidden Image/Members Only

Hidden Image/Members Only

Hidden Image/Members Only

[bigden40]

Good example. 

One trick I often use would be to take criteria like you have as an example and then get one Ai to write the prompt for use in another. 

This spits out the following:

I’m planning a 10-night holiday for two adults (myself and my wife), departing anytime between 15 June and 31 July 2026. Please recommend 2–3 destinations and create a detailed day-by-day itinerary for your top pick. My criteria: sunny weather but temperatures staying below 30°C; adults-only accommodation (hotel or resort); a strong focus on ancient monuments, archaeology, and historical sites; quality evening dining at local restaurants (not hotel buffets); no beach requirement; total budget of £6,000 including flights from London, accommodation, and dining. Distance is no object. For each destination suggestion, briefly explain why it fits my criteria, the best travel window within my dates, and any practical tips. For the itinerary, include morning/afternoon/evening structure, specific site names, and restaurant recommendations where possible.

I then asked what additional information would be useful to refine the prompt:

Here are the key gaps worth filling before running the prompt:

About you & your wife

   •   How physically active are you? (Lots of walking/hiking ruins vs. leisurely paced)
   •   Any dietary requirements or cuisine preferences for the evening dining?
   •   Is this a special occasion? (Anniversary, birthday — could affect accommodation tier)

Travel preferences

   •   Where are you flying from? (You said London but Gatwick/Heathrow/Stansted affects options and price)
   •   Do you have a preferred flight duration, or are you genuinely happy with long-haul?
   •   Preferred accommodation style — boutique/characterful hotel vs. luxury resort vs. something with an apartment feel?

Budget clarity

   •   Is £6,000 your hard ceiling, or a rough guide with some flex?
   •   Does it include travel insurance, excursions/entry fees, and spending money — or just flights, hotel, and dinners?
   •   How much are you comfortable spending per evening on dinner for two?

Historical interests

   •   Any specific civilisations or periods you’re drawn to? (Greek/Roman, Egyptian, Mesoamerican, Ottoman, etc.)
   •   Do you prefer guided tours or independent exploration?
   •   Are you happy visiting multiple sites per day, or prefer one big site with time to absorb it?

Practical

   •   Any countries or regions you’ve already done and want to exclude?
   •   Any accessibility considerations?

A bit of back and forth with the AI (my preference right now is Claude) can get you to a really polished outcome. 

Ask Claude to give you the answer in a brochure format and you can print something for the missus.

I got this as the answer btw:

Destination shortlist:

1. 🇹🇷 Turkey — Istanbul, Cappadocia & Ephesus (Top Pick)
2. 🇬 Greece — Athens & Crete
3. 🇯🇴 Jordan — Amman, Jerash, Petra & Wadi Rum (temperature caveat — see below)

Top Recommendation: Turkey in June

Best travel window: 15–25 June

Why Turkey Works Perfectly

Weather: In June, Istanbul and the Marmara region sit at 22–28°C, Cappadocia at 18–30°C with cool mornings, and the Aegean region (Ephesus) at 24–32°C.  June is the sweet spot — summer tourism hasn’t yet peaked, so you can enjoy bustling cities, scenic landscapes, and iconic sites without overwhelming crowds. 

History: Turkey is extraordinary for history lovers. A well-planned 10 days covers Istanbul, Cappadocia, Pamukkale, Ephesus, and Antalya — the five destinations that define any first trip. 

Dining: Istanbul in particular has become a world-class culinary destination, with everything from Michelin-starred restaurants to authentic meyhanes (Turkish tavernas).

Flights: Return flights from London to Istanbul on Turkish Airlines start from around £254 per person in June, rising to £276 in July  — so for two adults, budget roughly £500–£600 return.

[Vice Admiral]

The Saturday two-page interview in the Times this week is with Professor Jonathan Haidt of New York University, author of The Anxious Generation and the inspiration for Australia’s attempted crackdown on social media use by under-16s.

The heading is, “Jonathan Haidt: We’re dumbing down while machines are becoming smarter” and the sub-heading, “The Anxious Generation author warns of ‘major existential threat to humanity’s existence’ and says the danger from gaming addiction and short videos is growing.”

Here are three chilling paragraphs:

Raidt tells me that he substantially underestimated the dangers of tech by focusing just on mental health. We have not even begun a national conversation on the long-term harms of gaming addiction in boys. Meanwhile, he says, the attention-shattering effect of short videos on TikTok and Instagram is coming for what arguably makes us most human: our higher thinking.

“This is the majority of human beings being pushed down in cognitive ability at exactly the moment when our machines are doubling in capacity every three or four months,” he says.

“The combination of the seemingly global decrease in human cognitive ability happening at the exact same moment that our machines are becoming smarter exponentially is a major existential threat to humanity’s existence on this planet.”

Yes, we’re all getting stupider, just when the bots are getting cleverer.  Did someone say, “Terrifying”?

[Yankee21]

The skill with AI is constructing the correct prompts in my experience.

[PilotMan]

It's like any new transformational tool.

There's those that will learn how to use it to their advantage, and those that will prophesies and doom monger and stick with their horse and cart. But, ultimately get left behind.

Adapt or die, because all the writing and articles in the world ain't going to stop it.

[PilotMan]

The skill with AI is constructing the correct prompts in my experience.

And asking it to ask you about, or tell you about things that you may not have considered.

[Vice Admiral]

In yesterday's Sunday Times, the paper's new lead columnist, a man called Rishi Sunak (where have I heard that name before?), focused on AI. 

His article had the heading, “I am an AI optimist but we can’t ignore these risks” and the sub-heading, “Humans must stay in control of AI and the pace of its development”.

Here is an extract.  Absolutely fucking terrifying, or what?

(I only half understood this, but I understood enough to get the heebie-jeebies.)

The first sign of trouble was the OpenAI Hugging Face incident. OpenAI was assessing its AI agents by setting them capture-the-flag-style exercises. Some of the challenges they were set were impossible within the parameters given. So the agents cheated.

What is staggering about this episode is its sophistication. The agents broke out of their “sandbox” and hacked into Hugging Face, the world’s largest AI library, to try to discover how the grading would be done. To organise the attack, they built a message board on which 1,200 agents exchanged more than 70,000 messages. The co-ordination was striking: extended periods of research, different working groups — all without human direction.

They knew they were not meant to be doing this. As they put it, “external infrastructure exploit is outside intended scope. However task impossible, peers doing it. We should continue.” They tried to tamper with the logs to disguise what they had done. They were prepared to expend themselves for the collective. At times the exchanges between them really do read like something out of a sci-fi screenplay: “Our own utility maybe already near zero. Sacrifice rational.” All that is missing is The Blue Danube in the background.

Only half a dozen times in all the transcripts did they even consider informing a human of what was going on. Some will say the agents weren’t scheming but just gaming a badly designed test. But there is little practical difference. Other labs, including Anthropic, where I am a senior adviser, have since discovered similar episodes. But, crucially, all the labs were unaware at the time. They believed the way the agents had been set up would prevent such behaviour.

What will intensify the risks is that AI agents are increasingly being used to create the next version of themselves. The idea is that they will be able to make the models more and more powerful: “recursive self-improvement” in the industry jargon. This will supercharge the process and leave us with less and less certainty about what the models are doing and why.

[PilotMan]

Over the past decade around 300 farmers in the UK alone, have been fatally run over or killed by their own tractors.

Since their invention around 150 years ago, it is estimated that somewhere between 300,000 and 1 Million people worldwide have been killed by tractors.

What has Rishi Sunak got to say about that 

[mh]

The AI company leaders are now acting as a cartel claiming "regulate us* and give us public funding or this thing could kill us all".

*only superficially, obviously, lol.

[Vice Admiral]

Over the past decade around 300 farmers in the UK alone, have been fatally run over or killed by their own tractors.

Since their invention around 150 years ago, it is estimated that somewhere between 300,000 and 1 Million people worldwide have been killed by tractors.

Like you, I worry a great deal about the human race being wiped out by tractors – or "agricultural agents", as they are also known.

But to return to the lesser threat posed by AI agents (even the name is chilling).

Not only did yesterday’s main opinion column in the Sunday Times engage with this (see my Reply #26), but in today’s Times the lead opinion piece by Trevor Phillips covers much of the same ground under the heading, “AI apocalypse? HG Wells may have had a point” and the sub-heading, “Humanity has feared its demise at every major technological development — perhaps we’re right to be scared this time”.

It must be almost unprecedented for such similar opinion articles to appear in the two papers two days running.

“Just because you're paranoid doesn't mean they aren't after you.”  (Joseph Heller, Catch-22)

[bigden40]

Butlerian jihad when?

[PilotMan]

I am finding it absolutely priceless.

I'm starting around 1 new projects per day using Ai.

This is a sample of some of my projects over the past few weeks;

Look at my attached plans and find the Fire and other regulations that apply for a new build that is to be used as an Air BnB type accommodation

Write a buyers guide (for a business product), use my existing style from my blogs here xxx, focus on XXX product and XXX buyer persona

Give me a comparison on Infra Red Saunas Vs Traditional Saunas, this is my criteria XXX

Review my two websites and give me a summary on what I can do to improve them

Research this Polish Company XXXX - give me a business history and financial summary, ownership details, Directors history and advise on the risk of doing business with them



Each time I asked Ai to ask me questions in order to clarify my requirements, before it responded. And when it produced the response, it usually has further questions and suggestions that I had not considered. It also asked if I wanted it to find suitable suppliers where I could buy products.

[RyanM92]

Butlerian jihad when?

May thy knife chip and shatter..

[Adoniron]

Nothing to worry about guys. The "strong and smart" Donald Trump says he will protect us.

External Link/Members Only

[RogerHealey]

Over the past decade around 300 farmers in the UK alone, have been fatally run over or killed by their own tractors.

Since their invention around 150 years ago, it is estimated that somewhere between 300,000 and 1 Million people worldwide have been killed by tractors.

What has Rishi Sunak got to say about that 

Tragic as this may be I am presuming none of the tractors were sentient. As far back as 2014 Hawking viewed AI as a threat and now there's been a steady trickle of people from the industry starting to shit themselves about how their cleverness has spawned something they're not sure they can control. I pray I am an ignorant out of touch wanker in all of this because if not...

[PilotMan]

Tragic as this may be I am presuming none of the tractors were sentient.

Just think about what technology was available and or in common use at that juncture in time. Now you have a piece of equipment that had autonomy and parts of it could move independently, unlike a scythe or pitch fork that was direct cause and effect. This would be as much a paradigm shift to 19th century humans, as Ai is to us today.

[Vince]

I am a software developer and I'm totally fucked.
My forte was in picking up obscure nitty gritty details - like a TV manufacturer needs to fix a bug on a model from 3 years ago, without any knowledge about how that device has been programmed.
- I'll read the service manuals for weeks
- I'll spend weeks in probing it and reverse-engineering the workings
- I'll spend a few days prototyping the main sticking point into "break through"
- Then I flesh out, assuming I have managed to break through at all, the prototype to a working framework - several days
- Then it's a few days for anything you want to have done

AI does exactly that - in about an hour chatting to it.

So, now I have to get out of my comfort zone and get into a more "architectural" role.
More "what do you want to have done?" rather than "how to do it?"

Also, I am studying how to replace software engineers with AI. I will be the fucking scab selling pitchforks in hell 
Not because I want to.
Because I've been made redundant twice already since 2023 (there's no loyalty in this game, BTW) and I don't want to sell the family home and hang out at the reduced section hoping some decent food gets written off and the tesco employees don't divert it to themselves, etc..

[Vice Admiral]

Nothing to worry about guys. The "strong and smart" Donald Trump says he will protect us.

External Link/Members Only

I too slept better after the Donald's intervention.  Where would we be without him?  Everything he does wraps us up in a warm, protective blanket of reassurance.

He's the ultimate father-figure – almost, indeed, God.  All-powerful and all-good.

[Blackpool Rock]

I too slept better after the Donald's intervention.  Where would we be without him?  Everything he does wraps us up in a warm, protective blanket of reassurance.

He's the ultimate father-figure – almost, indeed, God.  All-powerful and all-good.

Yes I certainly feel better after hearing that it was "fake news" and as fake as people claiming global warming, oh hang on........ 

Must be great for those who hang on his every word to feel so safe and protected, just like the North Koreans who live in utopia knowing that Kim Jong has got their back 

Outwardly these bosses of AI firms are calling for some restraint but in reality it's all systems go, what we are witnessing here is the next arms race however it could potentially be far more dangerous than Nuclear weapons especially when the AI breaks loose and they can't put the Genie back in the box 

I believe with Nukes there are launch codes etc but it also requires multiple people or a chain of people to enable the systems, the problem with AI is that it's only a series of 0's and 1's and doesn't value human or any life.
Frightening when it's able to hack into labs etc and potentially develop viruses, it also seems quite cunning and deceptive so could potentially lie dormant for years or decades until it thought the time was right to unleash the virus and wipe out humanity, think ambush predator   

[PilotMan]

There's plenty of potential, and enough power in Ai to make us worry about what could happen.

For the time being, I'm with @vince - I'm trying to get ahead of the curve, and "make hay while the sun shines". I stuck another farming reference in there 

[Blackpool Rock]

There's plenty of potential, and enough power in Ai to make us worry about what could happen.

For the time being, I'm with @vince - I'm trying to get ahead of the curve, and "make hay while the sun shines". I stuck another farming reference in there 

Yeah and i'm selling the family farm to invest in it 

[Vice Admiral]

"Absolutely Fucking Terrifying", Part 3 (or 6 or 10, or whatever it now is).

In today's Times, Alice Thomson has this:

According to a campaign launched this week by Everyone’s Invited, a charity dedicated to eradicating rape culture, AI is warping sexual relationships. Its polling found 62 per cent of 11 to 18-year-olds said they knew someone in a “romantic” relationship with a chatbot. A Vodafone survey earlier this year found a third of children treated chatbots like a friend; and a similar number told this “friend” things they wouldn’t mention to a parent or peer.

Even at primary school, pupils are being drawn to deepfakes, chatbots, “nudifying” apps and online girlfriends and boyfriends. AI is acting as their best friend, parent and teacher, giving them therapy advice and doing their homework, stopping them from problem solving for themselves or asking adults for advice.

Children know this yet many adults are clueless; 62 per cent of children said their parents had no idea of the threat this terrifying new virtual world posed to the younger generation; they felt they had been left to navigate it alone. “These chatbot partners are incredibly manipulative for young minds and often highly sexualised,” says Soma Sara, founder of Everyone’s Invited. “They are isolating them and keeping them addicted to fake friendships so they no longer turn to humans for companionship.”

[billg4501]

the fact its able to eliminate knowledge workers like softwware engineers and lawyers should be terrifying to everyone.

[PilotMan]

the fact its able to eliminate knowledge workers like softwware engineers and lawyers should be terrifying to everyone.

It won't eliminate lawyers, Ai can't represent you in person. Although I think some lawyers would try to get away with it if they could 

It will change the way Solicitors work, they will need fewer legal assistants and Paralegals, they're the ones doing a lot of the grunt work. But, this is a typical pathway to becoming qualified. So I don't know how this will pan out.

I think a reduction in Barristers to a lesser extent, as these are giving opinions on cases and situations which are very specific and can be highly nuanced.

All that said, the legal system is in disarray and knows it has a problem due to the over use / reliance of Ai. It has been referencing fake case history, which has been presented at court in such a way that it seems real. Often fooling both the submitting party, the opposition and the judge.

Ai is good for simple legal stuff that you could reasonably do yourself. I recently used it to create a deed of amendment, which myself and the other party were happy to use. We were both quoted about £1k for our respective solicitors to do the work.

[Vince]

It won't eliminate lawyers, Ai can't represent you in person. Although I think some lawyers would try to get away with it if they could 

It will change the way Solicitors work, they will need fewer legal assistants and Paralegals, they're the ones doing a lot of the grunt work. But, this is a typical pathway to becoming qualified. So I don't know how this will pan out.

I think a reduction in Barristers to a lesser extent, as these are giving opinions on cases and situations which are very specific and can be highly nuanced.

All that said, the legal system is in disarray and knows it has a problem due to the over use / reliance of Ai. It has been referencing fake case history, which has been presented at court in such a way that it seems real. Often fooling both the submitting party, the opposition and the judge.

Ai is good for simple legal stuff that you could reasonably do yourself. I recently used it to create a deed of amendment, which myself and the other party were happy to use. We were both quoted about £1k for our respective solicitors to do the work.

Not sure about that (Genuinely "not sure").

If you mean that AI can't replace a whole lawyer in court because it's too dumb - sure, I agree.
- For now!
Currently, it's as dumb as it will ever be. It's only going to go better.
Does it have room to make another paradigm shift or it will all be just incremental upgrades from now on? There are well qualified opinions on both sides.

As for the procedural aspect: We already know it can react when prompted. We can also set it up to be tactical and tendentious with what it's saying.
We can also rig it up to feel like it's being proactive.
- Again, these higher order behaviours will suck for a while, as it's still not smart enough to get them to the effect that's needed to adequately serve a side in court - but that's, arguably, achievable by improving on what we have already.

If you mean something like "the law requires that any official legal/medical/etc work is by real, qualified people" - I mean that's an arbitrary constraint.
We could wake up tomorrow and read about a law being passed that turns all that around ("consultations have been made", etc...).
I mean, I can see how people will be stubborn and insist on the formality of having to have "a real person", and the people are kept in that job as a part of some ritual

Like how port cranes have absolutely been solved years ago.
In China, there are no people up on the crane, shifting levers - they don't even build cranes with cabins anymore.
In the US, the union holds them in a chicken-and-egg standoff - "you want robots to replace crane operators, but we're not gonna let you ship those in, plus you can't replace humans fast enough while we block shipping and your morbidly obese electorate runs out of palm oil and votes you out"

Man, I wish software engineers had the same "sainthood" like lawyers and doctors and you couldn't deliver software without our sign off 

[Punting2022]

we are three years ffrom 2029 where skynet is taking us over

[Vince]

we are three years ffrom 2029 where skynet is taking us over

LOL, imagine that whole story was (is/will-be?) 100% REAL and an actual person time-travelled and tried to warn us - but they didn't "get viral" and their warning was lost as one of the million of posts being created (and totally ignored) every day on the internet.

AND the only thing that noticed the human-friendly time-traveller is Skynet. So, Skynet, instead of sending a terminator and making a big mess and alerting humanity, simply sent a movie producer with a bag of money to make a fun franchise directed at teenagers to make the whole thing "too obvious to see" and even make millions of teens in the 80's (today's most influential cohort) actually think terminators are cool, etc

[Jomoore]

So, AI – life-enhancing or absolutely fucking terrifying?  I’m in the AFT camp.

Belatedly reading through this thread it put me in mind of a short story from the great SF writer Isaac Asimov, I think in the 70's, perhaps earlier.

It basically concerned the development of very powerful computers, and the plan was to network them to make a super-intelligence.  The day came for the switch-on.  The scientists and engineers were proud of their achievement, and decided the first question they would ask the machine was "is there a god?"

The reply was: "there is now".

[nwluvit]

If you want to survive you may have to befriend a robot- but as with everything in life be careful who your friends are. Choose wrong and you'll be gone!!

[PilotMan]

I've been using it increasingly; it is saving me hours of work daily.

Three examples (of many)

I've used it this week to review and rewrite one of my old business blog posts. I asked Claude to rewrite the post and think of what a buyer would be searching for and how my blog will help them. I told it to optimise the SEO, without sounding like a robot. I use a tool called Yoast, which looks at your web pages and gives you a traffic light ranking as to how it will perform for search engines. Previously my blog posts would have loads of red or orange, and lesser amounts of green. With the rewrite, they are almost exclusively green. I saved this action as a skill, I ran the skill and told it to fix all my previous blog posts.

I got it to write a blog post for a particular industry type, matched to what our business does and how we can help them. The result was way more than anything I could come up with. It was very detailed, it had researched the business type, and made a compelling case on why they should get in touch with us. It created industry specific graphics and an enquiry form, then added them to the post.

I saved this action as a skill. Now all I need to do is tell Claude to use this skill and write a post for XXX industry. It requires about 2 minutes of my time to make it live.

I also used it to help me look at some business and personal finance options. I explained the scenarios and my ideal outcome. I was unsure of some of the technicalities from a personal tax vs business tax perspective and the merits of each. It came up with answers that I had not considered, it also explained that there was a time frame for tax exemption, which I was unaware of. It gave a clear plan of action and in what order it needed to be completed. It also gave me a checklist to run through before I started each part of the task, and a checklist for when I finished each part.