Making a Smaller Community with ChatGPT « ipSpace.net blog
[ad_1]
I ought to be a good prompt engineer – just about every time I check with ChatGPT a little something seriously uncomplicated it spews out nonsense. This time I asked it to develop a modest network with four routers:
I have a community with 4 Cisco routers (A,B,C,D). They are related as abide by: A-B, B-C, A-D, D-C. Every single router has a loopback interface. Create router configurations that will outcome in A becoming ready to achieve loopback interfaces of all other routers.
Here’s what I obtained back again:
Here’s an case in point configuration for the four routers that should make it possible for Router A to access the loopback interfaces of all other routers:
It’s possible it’s just me, but I would expect an qualified to request a issue or two very first, like “are there any certain IP tackle you’d like to use, or is it all the identical to you”. Disregarding that bedside manners depth, let us inspect the configuration:
Router A
interface FastEthernet0/
ip deal with 192.168.1.1 255.255.255.
duplex car
pace car
!
interface Loopback0
ip handle 10...1 255.255.255.
!
router ospf 1
community 192.168.1. ...255 space
network 10... ...255 place
!
ip route ... ... 192.168.1.2
Router B
interface FastEthernet0/
ip address 192.168.1.2 255.255.255.
duplex vehicle
pace car
!
interface FastEthernet0/1
ip deal with 192.168.2.1 255.255.255.
duplex automobile
pace car
!
router ospf 1
community 192.168.1. ...255 spot
network 192.168.2. ...255 space
!
ip route ... ... 192.168.2.2
Router C
interface FastEthernet0/
ip deal with 192.168.2.2 255.255.255.
duplex auto
velocity auto
!
interface Loopback0
ip handle 10...3 255.255.255.
!
router ospf 1
community 192.168.2. ...255 region
community 10... ...255 spot
!
ip route ... ... 192.168.2.1
Router D
interface FastEthernet0/
ip address 192.168.3.1 255.255.255.
duplex car
pace car
!
interface Loopback0
ip tackle 10...2 255.255.255.
!
router ospf 1
network 192.168.3. ...255 spot
community 10... ...255 location
!
ip route ... ... 192.168.3.2
ChatGPT also described what it did, setting up with:
In this configuration, Router A is connected to Router B through FastEthernet0/ and to Router D through FastEthernet0/1. It has a loopback interface with IP address 10…1.
[…]
Determine out everything that’s mistaken with the above configurations just before going on )
How about:
- While ChatGPT mentioned the A ➔ D link, it is missing from the configuration of router A.
- Router D has a solitary interface that does not match anything else (due to the fact A ➔ D url is lacking).
- Router B has no loopback interface, so it is extremely hard to attain it 🤔
- Loopback interfaces have overlapping subnets (they are configured as /24s not as /32s)
Then we have the chain of unnecessary static default routes: A ➔ B, B ➔ C, C ➔ B, D ➔ A?. Charming )
But Hold out It Will Get Improved!
Of class, it will. GPT-4 does a substantially better job (many thanks once more to Andrew Yourtchenko for operating the query for me). Here’s what it produced for Router A:
It’s a key advancement in contrast to what ChatGPT created and very close to a ideal respond to:
- All the necessary inbound links are there
- Loopbacks have the predicted subnet masks
- P2P backlinks use lesser subnets
- OSPF is configured correctly, even though not optimally.
Even what ChatGPT created would be valuable if you choose it with a grain of salt as a starting off issue for a analysis project that would sooner or later end result in a managing community, but that’s not how the AI cheerleaders are marketing the GPT miracles.
Dependent on all I have noticed so far I are unable to rely on anything at all ChatGPT generates with no validating it, which suggests it could be an interesting absent-minded assistant, but I would even now have to know what I’m performing.
Disregarding the ontological dilemmas, I never think it is possible to give reliably correct responses devoid of comprehending what you’re chatting about. Massive language versions do a excellent occupation extracting the that means from a no cost-type question and creating a reply, but just due to the fact you consumed all the guides in the world doesn’t necessarily mean that you know or recognize what they are all about. Do not feel me? Browse a graduate textbook about quantum physics and notify me how that worked out.
Just one could also argue about the mysterious qualities of emergent conduct – when a model is substantial and complex plenty of, it could possibly exhibit unpredicted conduct, like seemingly knowing points. Does that remind you of how they explained to us OpenFlow would solve all networking complications?
Takeaway: As of March 2023, generic significant language product is almost certainly not the very best instrument to layout networks and develop community configurations. Let’s see what the upcoming week’s design delivers )
[ad_2]
Source backlink