Just like former computers of old, where prompts were in a C/ format, prompting LLMs (Large Language Models, or AI) is just as important. This is the number one reason LLMs give “incorrect data, information, or “hallucinate”.
Internet search bars were/are somewhere in the middle: too broad and you get everything. Too precise and you get nothing. I am going to provide an example on a proper query and show the results.
My prompt: Please explain, in paragraph form, the larger picture and ultimate goals of the CERN particle collider; for example, the “God Particle”, the reason for continuing to smash particles, what discoveries they have made so far, why it is shut down for 4 years, and the aim when it is turned on again in 2030. Please use real citations, any overviews of scientific articles that can be summarize, and any online scientific community societies of note.
Notice I stated what I wanted, how I wanted it given, the various aspects I want information on, and the types of research I want it to get it’s information from (no made up stuff) with citations and links (so I can double check accuracy. This is true for anything you do with an LLM. If you are looking for news, state the news sources you prefer. If you want unbiased news, state that and have it give voices from both sides. You can tell the LLM to remember your preferences.
Let’s see our answer: “The Large Hadron Collider (LHC) at CERN is designed to probe the deepest structure of reality by recreating conditions similar to those just after the Big Bang. Its “larger picture” is to test the Standard Model of particle physics, search for physics beyond it, and understand why the universe looks the way it does. The discovery of the Higgs boson in 2012—popularly nicknamed the “God Particle”—was its most famous achievement, confirming the mechanism that gives elementary particles mass. But the Higgs was never the end goal; it was the beginning of a deeper exploration into unanswered questions such as dark matter, matter–antimatter asymmetry, the nature of quark–gluon plasma, and the possibility of entirely new particles or forces.
“Why continue smashing particles?
Particle collisions at extreme energies allow physicists to observe rare phenomena that cannot be produced in any other environment. Over its first three runs (2010–2026), the LHC produced:
- 54 million billion proton collisions for ATLAS and CMS, enabling unprecedented precision measurements.
- ~300 billion heavy-ion collisions, revealing properties of quark–gluon plasma—the primordial state of matter after the Big Bang.
- Discovery of more than 85 hadrons, including exotic combinations of quarks.
- Thousands of peer‑reviewed papers (~4500), advancing astrophysics, nuclear physics, and cosmology.
These collisions also help set exclusion limits—ruling out where new particles aren’t—which is essential for narrowing down theories of dark matter and other beyond‑Standard‑Model physics.” (You might say, “that’s in bullet points; you requested paragraphs”. When I first started working with my LLM, I asked it to remember to break tasks down for me in small chunks with bullet points. It has a long memory.)
“Why the LHC is shut down for four years (2026–2030)
The shutdown, called Long Shutdown 3 (LS3), is a planned, multi‑year overhaul to prepare for the High‑Luminosity LHC (HiLumi LHC). This is not a repair but a metamorphosis. CERN is:
- Dismantling 1.2 km of the accelerator to install new superconducting magnets and cryogenic systems.
- Replacing key focusing magnets (“inner triplets”) with niobium‑tin superconducting magnets capable of 11.3‑tesla fields—40% stronger than current ones.
- Upgrading detectors across ATLAS, CMS, ALICE, and LHCb to handle vastly higher collision rates.
- Renovating other CERN facilities (ISOLDE, North Area Hall) and consolidating infrastructure.
These shutdowns are essential because the LHC operates near absolute zero; maintenance and upgrades must be done during long, carefully planned warm‑up periods.”
It has given me quite a lengthy response which many likely do not want to read (but if you do, let me know in the comments!), so I will add that we can refine our query -like I will do later- such as this: “Keeping all of this in mind, let’s look at France and why they have a center looking to “Capture the sun in a bottle” in 2030.” It will hold the current conversation, your preferences, the citations, and move to include the next piece as well as tell you how or if they are connected.
I hope this was useful! ~Professor
Leave a comment