You don't need to be an 'investor' to invest in Singletrack: 6 days left: 95% of target - Find out more
Just messing around with it to see what comes out!
You said:
Is schwalbe tacky Chan and big Betty a good combination of tyres on a 29er Enduro bike in the peak district
ChatGPT said:
Yes, Schwalbe Tacky Chan (front) and Big Betty (rear) is a very good tyre combination for a 29er Enduro bike in the Peak District, especially if you're riding technical, rocky descents and want aggressive grip and control.Why this combo works well:
✅ Tacky Chan Front
Designed for maximum cornering grip and braking traction in aggressive terrain.The Addix Soft or Ultra Soft compound offers great damping and bite, ideal for wet gritstone, roots, and loose rock common in the Peaks.
Works well in all but the stickiest clay-type mud – great for year-round use in the UK.
✅ Big Betty Rear
A fast-rolling, supportive rear tyre with plenty of braking edges.Good for technical descents and chunky climbs – perfect for the steeper Peak District routes (e.g., Jacob’s Ladder, Cut Gate).
Pairs well with the Tacky Chan to balance front-end grip with rear support and stability.
Suggested setup tips:
Casing: Choose Super Trail or Super Gravity casings depending on how gnarly your riding is (Super Gravity if you do big hits or uplift days).Compound: Soft (front) + Soft/Speedgrip (rear) is a good match for traction and rolling speed balance.
Pressure: 20–24 psi front, 23–28 psi rear (tubeless), adjusted to your weight and terrain.
Alternatives worth considering:
Magic Mary front for slightly more all-conditions versatility.Hans Dampf rear if you want something a bit faster rolling for longer rides.
But overall, Tacky Chan + Big Betty is an aggressive, modern Enduro setup well-suited for the technical, rocky, and often wet terrain of the Peak District.
Let me know your riding style or preferred routes and I can refine the advice further.
You said:
What pressure for a 77kg rider using inserts
ChatGPT said:
For a 77 kg rider using tyre inserts (e.g. CushCore, Rimpact, etc.) on a 29er Enduro bike with Schwalbe Tacky Chan (front) and Big Betty (rear), here are recommended tubeless tyre pressures:🔧 Recommended Tyre Pressures with Inserts:
Wheel Pressure Range Typical Starting Point
Front (Tacky Chan) 20–22 psi (1.4–1.5 bar) 21 psi
Rear (Big Betty) 22–25 psi (1.5–1.7 bar) 23.5–24 psi
🧠 Why this works:
Inserts add sidewall support, so you can run slightly lower pressure without risking burping or rim strikes.Front: Slightly lower pressure for better grip and compliance on roots and rocks.
Rear: Higher pressure for support on landings and improved rolling resistance, especially over rough terrain.
🔧 Adjust Based On:
Terrain: Chunky descents → go +0.5 psi. Wet, loose → go −0.5 psi.Riding style: Aggressive riders might prefer a touch more pressure to avoid tyre squirm.
Tyre casing: If you're running Super Trail, stay within the above range. If Super Gravity, you can go slightly lower due to stiffer sidewalls.
Final Tip:
After your first ride on this setup, check rim contact, tyre squirm, and grip to dial in the pressures. A digital gauge helps with fine-tuning (even 0.5 psi makes a difference).Let me know which inserts you're running if you want ultra-specific figures.
You said:
Cushcore
ChatGPT said:
Great — with CushCore inserts, you can safely run lower pressures than usual, thanks to the excellent rim protection and sidewall support they provide. This boosts traction, damping, and cornering stability, which is ideal for the Peak District’s mix of rocky tech and fast chunder.✅ Recommended Tyre Pressures for 77 kg Rider with CushCore (29" Enduro setup)
Wheel Tyre Recommended Pressure
Front Tacky Chan (Soft) 20–21 psi
Rear Big Betty (Soft or SpeedGrip) 22–23 psi
📌 Suggested starting point:
Front: 20.5 psi
Rear: 22.5 psi🔧 Tuning Tips with CushCore:
Lower limit: You can go as low as 19 psi front / 21 psi rear if conditions are wet or you want max grip — just monitor for squirm or casing roll.Higher limit: For bike park days or sharp Peak rock gardens (e.g. Ladybower slabs, Cut Gate), bump up to 21.5 front / 23.5 rear for more support.
Feel: With CushCore, tyres ride slightly “firmer” at the same pressure, so don’t be afraid to dip below your usual range.
Bonus:
If you're running Super Gravity casings, you can go even lower — 19 front / 21 rear is totally reasonable for gnarly days.For Super Trail, stick to the ranges above to maintain sidewall integrity.
Let me know if you’re riding flats or clips, or if you want to tune this setup for race vs. all-day epics.
You said:
Flats
ChatGPT said:
Perfect — riding flats means you’ll benefit even more from lower pressures and better tyre feel, but you also need to avoid tyre squirm or rim strikes on hard hits or compressions.🎯 Updated CushCore Pressure Recommendations (Flats, 77 kg rider):
Wheel Tyre Suggested Pressure
Front Tacky Chan (Soft) 20 psi
Rear Big Betty (Soft/SpeedGrip) 22 psi
🛠️ Minimum safe limit with CushCore + Flats:
Front: 19 psi (if super slick/wet and you want all grip)
Rear: 21 psi (any lower and you might feel vague when cornering hard)🧠 Why slightly higher in the rear:
Rear tyre gets more abuse — braking, square hits, pedalling.Stability over rock gardens and slabs (e.g. Stanage, Win Hill) needs a bit more pressure to prevent tyre squirm or sidewall fold.
👣 Why this works for flats:
Lower front pressure gives better feedback through the pedals.More grip = more confidence when weighting/unweighting the bike.
Reduces harshness through the feet on rough terrain.
Final Tip:
If you do a fast, rocky descent and feel any rim buzz or excessive roll, bump up 0.5 psi and re-test. CushCore makes it easy to dial in that sweet spot.Let me know if you also want suspension setup tips to match this tyre setup.
You said:
Yes please
ChatGPT said:
Excellent! Let’s dial in your suspension setup to match your CushCore-equipped, Schwalbe Tacky Chan + Big Betty tyre combo on flats, for Enduro riding in the Peak District.You’re 77 kg (rider weight, not including gear), and likely riding a ~160mm travel 29er. Here’s how to get your suspension working with your tyres for maximum grip, comfort, and control.
🔧 SUSPENSION SETUP GUIDE (77 kg rider with CushCore, riding flats)
📌 General Priorities:
Grip and tracking on technical rocks, slabs, and loose descents.Support for flat-pedal body position (more reliance on suspension for traction).
Balanced front-rear ride feel.
🔩 FRONT FORK SETUP (e.g. 160–170mm)
Setting Target Value (starting point)
Sag 15–20% (about 25–30mm on 160mm fork)
Air Pressure Check fork chart for your model (~75–80 psi for 77 kg)
Rebound Start mid-range, then slow slightly (fewer clicks from slowest)
Low-Speed Compression (LSC) 1–3 clicks from open (tune for brake dive and support)
High-Speed Compression (HSC) Open or 1–2 clicks in for rocky hits/slabs
✅ Goal: Front should stay planted in rough sections without diving too easily. CushCore + 20 psi front tyre already gives good compliance, so don’t overdamp.🔩 REAR SHOCK SETUP (e.g. 150–160mm)
Setting Target Value
Sag 27–30% (Enduro = more sag for traction)
Air Pressure Adjust to get correct sag (likely ~180–190 psi for 77 kg)
Rebound Mid-fast range (helps rear track terrain on flats)
Low-Speed Compression 1–3 clicks from open for pedal support
High-Speed Compression Open or 1–2 clicks in for rock gardens, depending on frame kinematics
✅ Goal: Rear should track terrain well and give grip under load. With CushCore + 22 psi rear, you’ll already have damping from the tyre insert — so avoid overdamping.🔁 SETUP TIPS
Balance: Do a bounce test — bike should compress evenly front to rear.Trail test: If front feels harsh or nervous, reduce HSC or drop 2 psi. If rear feels dead or wallowy, increase rebound speed or add a click of LSC.
Use volume spacers if you bottom out too easily:
Fork: Add 1 spacer if you use full travel too often.
Shock: Add 1 token to increase midstroke support.
🧪 Fine-Tuning Loop:
Start with the above base setup.Ride a trail you know well (rocky, steep, with corners).
Adjust 1 thing at a time, note the effect.
Aim for max grip without sacrificing control or pedalling support.
Would you like me to create a custom tuning chart (PDF or printout) to take on the trail? Or recommend specific settings for your suspension model (e.g. Fox 36, RockShox Lyrik, etc.)?
You said:
I have a super deluxe ultimate with megneg Aircan on a bird am9 and a lyrik ultimate 170mm fork
ChatGPT said:
Update memory? “Rides a Bird AM9 with a RockShox Lyrik Ultimate 17...”Yes
No
Perfect — that’s a very capable setup for aggressive Peak District riding. With the Bird AM9, Super Deluxe Ultimate + MegNeg, and Lyrik Ultimate 170mm, you’ve got a highly tunable platform. Here’s a detailed setup guide specifically for your rig and weight (77 kg), matched to your CushCore/tyre setup and riding flats.🎯 BASELINE SETUP: Bird AM9 (Super Deluxe Ultimate MegNeg + Lyrik Ultimate 170)
🔧 REAR SUSPENSION — Super Deluxe Ultimate with MegNeg
Setting Recommendation
Rider weight 77 kg (170 lb)
Sag (with MegNeg) 32–35% (target more sag for grip)
Air Pressure ~180–190 psi (start at 185 and adjust for sag)
Volume Spacers 2–3 MegNeg bands (tune support; fewer = plusher)
Rebound 7–9 clicks from closed (slow) (count from full slow)
Low-Speed Compression 2–3 clicks in from open
High-Speed Compression Fully open or 1 click in
📌 The MegNeg increases midstroke and end-stroke support, so you can run more sag without losing pedal support or ramp. Perfect for flats + grip-heavy riding.🔧 FRONT SUSPENSION — Lyrik Ultimate 170mm (Charger 2.1/3 RC2)
Setting Recommendation
Sag 15–20% (≈25–30 mm on 170 mm)
Air Pressure ~75–78 psi
Tokens 2 tokens for balanced support & ramp
Rebound 8–10 clicks from closed (slow)
Low-Speed Compression (LSC) 3 clicks in from open
High-Speed Compression (HSC) Open or 1–2 clicks in
📌 The Lyrik tracks really well with CushCore + 20 psi up front. These settings will keep it supple off the top while resisting harsh bottom-outs.🧪 Fine-Tuning Suggestions
On-Trail Adjustments Checklist:
Symptom Adjustment
Harsh chatter on rocks Soften HSC or reduce fork pressure by 2 psi
Rear feels dead/pogos Speed up rebound 1–2 clicks
Bottoming too often Add 1 token or increase HSC (fork) / add MegNeg band (shock)
Too much pedal bob Add LSC rear 1–2 clicks
Front dives under braking Add LSC on fork 1–2 clicks
🗺️ Peak District Specific Notes:
On tech like Cut Gate, Jacob’s Ladder, or Lady Canning’s Black, you’ll want that extra midstroke support from the MegNeg and fork tokens.For slabby riding (Stanage, Houndkirk), keep HSC open and rely on your inserts + grip setup to absorb hits smoothly.
Would you like a printable trail tuning worksheet or a chart to log settings and trail feedback? I can generate one tailored to this setup.
You said:
The lyrik is also running a trutune insert
ChatGPT said:
Updated saved memory
Thanks — with the TruTune insert in your Lyrik Ultimate, your fork gains a more linear, coil-like feel in the midstroke with better small bump sensitivity and support, without needing excessive air pressure or tokens.Here’s how this changes your fork setup:
🔧 Updated Lyrik Ultimate Setup (with TruTune insert)
Setting Recommendation
Air Pressure ~72–75 psi (slightly lower than without TruTune)
Tokens 1 token max, or none — TruTune adds natural ramp-up
Sag 17–20% (~28–34 mm)
Rebound 8–10 clicks from closed (slow)
LSC 2–3 clicks from open
HSC Open or 1 click in (TruTune handles midstroke support)
📌 With TruTune + CushCore, you now have a very supple front end that won’t dive or wallow under load — ideal for fast rocky chutes, slabs, and flat-pedal cornering.🔧 On-Trail Feel:
Better grip in rocky turns and off-camber rootsMidstroke support lets you push into the fork without spiking
Feels a bit firmer at sag, but more controlled deeper in the travel
🧪 Fork tuning advice with TruTune:
Avoid over-damping — keep HSC and LSC on the open side.Drop air pressure by ~3–5 psi vs. stock setup.
If it ever feels harsh, check that you don’t have too many tokens, or your pressure hasn’t crept up.
Would you like me to generate a suspension + tyre + insert tuning card (printable or digital) to use during trail testing or events? I can also include suggestions for wet/dry adjustments.
I can't be bothered to read all that. Maybe I need some sort of AI assistant to read the forum for me and filter out screeds of cut and paste AI generated bilge.
TBF that's pretty impressive knowledge
AI is going to ruin a lot of things, for all the good it may do I think the negatives will outweigh them, starting with the insane amount of energy it will consume.
I suspect you could have found out most of what it replied yourself, or better still experimented and worked it out yourself?
The point being that this produces useful info for the majority of people who cannot be arsed to do things for themselves and want answers ASAP. Also super specific to criteria as opposed to a dude on a ht on the South Downs with a long beard giving his two penneth lol
Just remember that's all basically copy and paste from MTB websites that now don't get the traffic and associated ad revenue.
Governments need to figure out how to make sure the output of AI is still crediting the original creators of the information, but there's so many other big things to deal with I can't see it happening.
Also, wouldn't be surprised if it said yes rocket ron is a good choice too, with an explanation that sounds perfectly reasonable.
I can't be bothered to read all that. Maybe I need some sort of AI assistant to read the forum for me and filter out screeds of cut and paste AI generated bilge.
You knew from the thread title it would.trigger you but still chose to be triggered. Maybe use chatgpt for CBT advice or even chatcbt?
😯
I read 80% of the report. Seems like a place to start with bike setup if you have never ridden somewhere before.
Did it produce a chart?
AI can be useful for sure.👍
But it's quite important to maintain our own knowledge/intelligence and to keep our own dialogues going in forums like this, and in person... as well as try things for ourselves.
With any powerful new tech (that's initially free), there's either potential for future £ fees or significant influence by the £ highest bidders to keep it 'free'.
The money to develop and maintain social media, AI, etc has to come from somewhere - so always ask the £ question vs. "what's in it for me in the long term".
Would I use it as a tool? Yes.
Would I rely on it? No.
I asked it why a carbon frame wheel alignment was off and it said the frame was scrap and I shouldn't ride it, structural integrity was compromised.
I thought about it, realised a carbon frame can't just bend like steel and there'd likely be visible damage somewhere. Turned out it had the wrong dropout fitted, frame was fine.
So don't trust everything it says. It has the same level of confidence regardless of whether its right or wrong.
The fact people are integrating this into government and defence is very worrying!
I also don't get the business model at all. The actual LLM has become commoditised, you can just swap one for another, or even run it locally for free.
The money seems to be in the applciation layer that integrates it? Seems to be same for autonomous cars, what Uber is doing makes sense, they can just swap out one supplier like Wayve for another, like Waymo, or even go back to human drivers.
ChatGPT is an LLM, NOT an AI - an LLM is a Large Language Model. It does not know things, it matches phrases and words and uses those to scan for likely answers. If you have enough "sources" saying the same thing, then your "answer" could well have been "No, schwalbe tacky Chan and big Betty are not good combination of tyres on a 29er Enduro bike in the peak district, you should use the skin of newborn children doused in petrol and set alight so your wheels are on fire".
At the moment, you're getting "good" answers becuase all the sources that have been scanned have a similar response. As "AI" progresses, it's beginning to feed on itself and user other AI LLM "answers" as a suitable source, and the problem is that if an LLM does not find the answer, it will make one up. If that is then taken as an answer, we go into a degenerative loop.
Current AI is not the answer to the problem. Current AI is the problem
But it's quite important to maintain our own knowledge/intelligence and to keep our own dialogues going in forums like this, and in person... as well as try things for ourselves.
This in spades for me. Chat GPT and the like will dumb down the human race....further.
Just remember that's all basically copy and paste from MTB websites that now don't get the traffic and associated ad revenue.Unfortunately it's also CnP from places like Reddit where a good percentage of the content is utter garbage.
Also, wouldn't be surprised if it said yes rocket ron is a good choice too, with an explanation that sounds perfectly reasonable.Which is what we're finding with AI stuff like this at work. It'll slip some utter crap in with a perfectly logical sounding explanation, unless you know what you're actually looking for/at.
The point being that this produces useful info for the majority of peopleIt produces a lot of unqualified word salad. That sounds shiny. You need to be able to qualify it.
Much like what Tory politicians say actually.
I can't be bothered to read all that. Maybe I need some sort of AI assistant to read the forum for me and filter out screeds of cut and paste AI generated bilge.
You knew from the thread title it would.trigger you but still chose to be triggered. Maybe use chatgpt for CBT advice or even chatcbt?
Because dumping a massive output from ChatGPT and expecting everyone else to decipher it and what point you are trying to make is lazy and has been done before. What's interesting are your thoughts which to be fair you did provide a few posts down.
It's stealing our jobs!
Well it didn't say: "No, the Betty is average at best and you really need the new radial carcass tyres if you're going Schwalbe", and then argue with itself that Conti or Maxxis are actually far better. Or say the Lyrik is utter garbage and only a Mezzer will do.
So our jobs are safe for now.
This from Katie Mack sums it up well for me:
Which doesn't mean LLMs can't be useful but just that it's important to understand their limitations. One of them is that the sheer volume of text they can produce quickly can make it hard to assess the output well. The I struggle with long sentences replies to the OP kind of demonstrate this - I imagine someone with a detailed knowledge of MTB tyres might find fault with the info provided but who honestly could be arsed to do that?
Unfortunately it's also CnP from places like Reddit where a good percentage of the content is utter garbage.
Including deliberately seeded garbage. As soon as Reddit announced it was allowing its content to be fed into LLMs people started spamming it with posts to try and influence the "answers" those models give.
It's the "can't avoid" nature of LLM "answers" that is getting to me... Meta have put them first and foremost in so many locations... Google search, WhatsApp... so you see the nonsense before human quotes/links all the time. I had food poisoning... looked up what to eat and drink first once better... the "answer" generated by Google, that took up most of the screen on my phone, said to drink Gatorade... following the first 5 natural results/links below the "answer" they all said NOT to drink Gatorade. So Gatorade pops up in all the answers written or curated by humans, and the LLM text also mentions drinking Gatorade... but it didn't really have the knowledge about whether and why drinking the product was the right or wrong advice. Poor advice. Unless you're PepsiCo, I suppose.
I imagine someone with a detailed knowledge of MTB tyres might find fault with the info provided but who honestly could be arsed to do that?
Oh come on, you're not new here - this is what we live for 😉
Don't forget to check it's not making it up as it goes along...
We welcome our new AI overlords! 🖐️
I also don't get the business model at all.
I think an interesting aspect of the buisness model was revealed when new, cheap LLMs from china appeared and everyone said 'but wait how can this be? AI is really expensive theres no way China can be doing it for a fraction of the price without cheating' and at the same time you see Nvidia have been charging a 90% margin on the chips they've been supplying to all the AI companies 🙂
This in spades for me. Chat GPT and the like will dumb down the human race....further.
I dunno. We could leave it to handle the "what tyre" threads which it ably can, and get back to reading Jane Austen...
Pretty sure I had an email about a case that got thrown out and the lawyer censured because two of the four previous cases they referenced in their submission turned out to be made up by whatever AI tool they used to research. 🙂
Pretty sure I had an email about a case that got thrown out and the lawyer censured because two of the four previous cases they referenced in their submission turned out to be made up by whatever AI tool they used to research. 🙂
yeah I remember seeing something about that too
Just wait till it gets hacked by a superior AI machine, then only feeds you suggestions that have been approved my its masters.
A life of dull with no surprises awaits 😉 🤣 🤣 🤣
TBF that's pretty impressive knowledge
Is it?
I mean it has provided answers by sifting through all of humanity's accumulated online waffle on a given topic and then condensed it into an answer to a series of reasonably simple questions. You could have gotten to the same (or possibly 'better') answers with a bit of googling and sifting of your own, and you might have a better understanding of the answers had you forced yourself to actually research the topic, it just wouldn't have been as rapid.
To me LLMs are just search engines with a filter so our puny human brains aren't overwhelmed by detail or nuance.
It's kind of the inverse of "Impressive Knowledge" People are using these tools to reduce their workload for basic (but time consuming) tasks, but they often end up filtering out most/all understanding and who is actually checking the answers?
And it creates the same false expectation that many technologies seem to have over the years, it will "Save us work" and make us all "more productive". What really ends up happening is that more is expected of us in the same amount of time. Are people who heavily augment their day to day work with "AI" doing 6 hour days or 4 day weeks? No they're simply cramming more "work" (output) into the same ~40 hour(+) weeks, or working longer to keep up with the slop their new toys churn out, ultimately upping the ante on stress, workload and potential for errors...
I'm increasingly of the opinion that "AI" isn't really helping humanity at all, just herding us all towards a faster demise while we try to keep up with increasingly wild KPIs. Of course once the bubble bursts and our Billionaire overlords realise they're spunking lots of money for minimal returns, the hype will slow down... hopefully.
You could have gotten to the same (or possibly 'better') answers with a bit of googling and sifting of your own,
Or possibly worse. After a couple of hours rather than seconds.
I don't buy the "it's glorified auto correct therefore it's not really clever" line of argument. Judge by what it comes out with rather than how it got there.
The couple of people I know who have worked and interacted with it on technical level are unnerved by the fact they can't actually follow how it works. We don't really know how human intelligence (oxymoron?) works either but do know it's massively fallible . Probably why I'm still running minions
opposed to a dude on a ht on the South Downs with a long beard giving his two penneth lol
Oi.....Some of us on the SD have short beards and short travel FS XC rockets 😀
But which bits of that reply was ChatGPT making up or hallucinating? Colleague writing a clinical paper the other day asked ChatGPT out of interest, and it was making up references with good sounding titles and journal DOI numbers. RFK Jr presented his new vaccine case report the other day, and a senator with an MD took him to task because half the refernces his recommendations were based on were pulled out of thin air - and RFK was changing national vaccine policy off the back of it.
I suspect you could have found out most of what it replied yourself
And what would the carbon footprint of that research be?
ChatGPT/LLM AI is just another step in a long line of steps
Google was a game changer, a mini revolution being able to search all of the internets in one go.
IIRC the predecesers like Jeeves and Yahoo just never really hit the spot.
A large langauge model is just a development of the search engine.
Of course it needs sense checking, but you dont take the first google link as gospel, do you?
The couple of people I know who have worked and interacted with it on technical level are unnerved by the fact they can't actually follow how it works.
I'm not sure that's a good thing though?
It's firmly established that it makes things up, and it's not always clear what it's made up.
I write and edit things for a living. Occasionally AI tools can save me time and give me ideas, but ask it to do any substantial work and you'll often spend more time fact-checking it than if you'd just written the thing yourself.
For those of you that jump to AI for everything, these are worth reading:
https://www.nytimes.com/2025/06/13/technology/chatgpt-ai-chatbots-conspiracies.html
And
https://time.com/7295195/ai-chatgpt-google-learning-school/
I also don't get the business model at all. The actual LLM has become commoditised, you can just swap one for another, or even run it locally for free.
The current business model is roughly thus:
1. Create LLM
2. Spend billions and billions on data centres, compute and GPUs, to try and improve the LLM, and offer it very cheaply to users
3. ????
4. Profit
We are currently at step 2. The major AI companies are losing many billions per year, and are dependent on a constant flow of venture capital so they won't go bankrupt.
The reason you constantly hear the CEOs of the AI companies in the news claiming that LLMs will replace all jobs in the next week is because they desperately need more funding, while they figure out what step 3 is and how they might actually achieve it. If they can convince the world - and more importantly the likes of Microsoft and the big VC firms - that step 3 is all in hand, then they might get a few more months' funding.
ChatGPT said:
Perfect — riding flats means you’ll benefit even more from lower pressures and better tyre feel, but you also need to avoid tyre squirm or rim strikes on hard hits or compressions.
Is that true - I've no idea but can't see why
Air Pressure ~72–75 psi (slightly lower than without TruTune)
I thought general recommendation was to RAISE pressure a bit
— TruTune adds natural ramp-up
I thought the reverse was true
I wouldn’t know. I refuse to use that AI crap
Judge by what it comes out with rather than how it got there
Of course it needs sense checking, but you dont take the first google link as gospel, do you?
Here's the rub. They make blindingly obvious mistakes all the time. We fix the obvious mistakes and pass on the rest of the unobvious mistakes. LLMs combine pools of knowledge and provide an incorrect summary of that knowledge, and then we accept those summaries (even though we've all seen that the process used to collate them is totally flawed).
I asked chatgpt a while ago how thick TRP brake discs were. It told me that generally, Mtb brakes are between 160 and 220mm.
It then told me that on average MTB brake discs are between 55 and 65mm thick.
A d there's me worrying about the eight of a tyre insert...
I just asked it about my suspension settings. It told me to add 3 tokens and 6 clicks of HSC.
It's a Manitou fork that doesn't use tokens as a means of adjustment and only has 4 clicks of HSC.
OPs results sound really impressive, the detail is excellent but I do question the validity.
As mentioned above, it's suggesting to decrease air pressure after fitting the TruTune. The whole point of TruTune is that it let's you run higher pressure for mid stroke support whilst magically tuning out some of the progression. It doesn't 'add natural ramp up', it reduces it.
The way I think of them is that they know all the rules that relate words. For instance ’up’ is the opposite of ‘down’. They just don’t experience any of the things described by the words. They can equally describe pain and pleasure. It’s all just a vast set of linguistic relationships.
In the real world you have to remember that if you don’t keep things rubber side down, pain will not just be the opposite of pleasure, it can end up in A&E. Hence trying, testing and learning need to be done for real … fine words are not necessarily your friends
So don't trust everything it says. It has the same level of confidence regardless of whether its right or wrong.
The fact people are integrating this into government and defence is very worrying!
It's being pushed at us (local government) all the bloody time. Teams meetings recorded by "AI Minute Taker", emails and document summary by CoPilot...
It all *sounds* impressive. It's like listening to a very competent public speaker or thinking that you're reading a scholarly book.
But it needs a lot of knowledge of the subject to realise that most of what it churns out is bollocks.
You could have gotten to the same (or possibly 'better') answers with a bit of googling and sifting of your own, and you might have a better understanding of the answers had you forced yourself to actually research the topic, it just wouldn't have been as rapid.I see it as a step down from using Wikipedia as a source. With Wikipedia you get a list of useful links to actual sources (usually), which helps with a *very* basic start to actual research. Most of the LLM/AI/Chat GPT/copilot models you get two pages of word salad and some links that may or may not link to anything useful. Or that might just be made up!
I see it as a step down from using Wikipedia as a source. With Wikipedia you get a list of useful links to actual sources (usually), which helps with a *very* basic start to actual research. Most of the LLM/AI/Chat GPT/copilot models you get two pages of word salad and some links that may or may not link to anything useful
But I used to love uncyclopedia !
It all *sounds* impressive. It's like listening to a very competent public speaker or thinking that you're reading a scholarly book.
Leave Boris Johnson out of this
I’m tempted to start putting completely wrong answers on all forums and social media posts to try to help them get wrong answers. It’s dumbing down people. My cousin is a lecturer at a University the amount of ChapGPT written essays is off the charts. I’m a paramedic and the amount of students we have who are openly telling us they just use it to write their essays is worrying.
I’m tempted to start putting completely wrong answers on all forums and social media posts to try to help them get wrong answers.
There are many people already ahead of you, it is called data poisoning. They are probably using AI in some form to help them do it...
I'd be curious to know how much work and progress is being done by the AI companies to address this.
Sure they're all LLMs, made for genuine seeming text and not accuracy, but in reality companies are pushing these tools as search engines and helpful sources of information, surely it's in their best interests to make the output as accurate as possible. I would be very surprised to learn that they are "pure" LLM with no additional processing to try and improve accuracy and (meaningful) relevance.
They're insanely short-sighted if they're not just as worried as everyone else is about the self-poisoning tendencies of widespread AI usage. Maybe they really just don't care 🤷♂️
The reason you constantly hear the CEOs of the AI companies in the news claiming that LLMs will replace all jobs in the next week is because they desperately need more funding,
Hasnt there been a notable drop in recruitment of graduate jobs over the past year?
You need someone knowledgeable and experienced to sense check chatGPT output, but you do with Graduates too.
and it will only get better (i believe)
Thats potentially a bit worrying for the entry level job market.
It fills a lot of holes in my programming skills. Im not a coder or a programmer, but i know enough to know what to ask ChatGPT, and ChatGPT knows enough to be able to turn my prompts into some serviceable Python or SQL.
Or more likely, i can mash together some code rapidly, without worrying too much about syntax and specifics, and throw it into ChatGPT and say "fix this".
its really good at that, in my experience. Its playing to its strengths though. There is no leeway in script syntax, so theres no conflicting opinions about how it should work knocking around on the internet to confuse it. (as long as it can filter out the posts on forums titled "my script doesnt work"
im not programming CERN, but it has certainly vastly reduced the liklihood we would employ a programmer to fill that skill gap.
tools as search engines and helpful sources of information, surely it's in their best interests to make the output as accurate as possible
I think we’ll see more companies leveraging the data they have to provide better quality answers and let you choose the appropriate tool for the job - for a cost. For example, if you want to know some background on a science topic you’ll probably get a better answer (compared to chatGPT) from Scopus AI, which is trained using abstracts from peer reviewed articles in the Scopus database (owned by Elsevier), without pollution from the detritus elsewhere on the internet. It will also give you the (real) citations it has based the responses on.