DEV Community: Lazar Tomić

The Gap Between AI Hype and Scientific Reality

Lazar Tomić — Sun, 28 Jun 2026 04:08:58 +0000

Where is the reality then?

A few years ago, my understanding of artificial intelligence came from the same places as most people’s: headlines, social media, YouTube videos, and the occasional viral breakthrough. AI seemed almost magical. Every week, there was a new claim about machines replacing jobs, achieving human-level intelligence, or transforming entire industries.

Then, I had the opportunity to spend time in a real research environment, for a matter of fact, one of the most important European scientific facilities: GSI/FAIR.

What I found out there was not disappointing; it was rather a turning point and an eye-opening experience.

Most of my time at high school, peers spoke about it when it came to homework and learning… but as the models advanced, the dependency on them became more and more noticeable.

The AI presented to the public is often a finished product: polished demos, impressive benchmarks, and bold predictions.

Research, however, looks very different. It is slower, more rigorous, and far less predictable. Progress is measured in months and years rather than days. Experiments fail. Models behave unexpectedly. Hardware limitations become impossible to ignore.

As a high school student, I expected to see groundbreaking algorithms everywhere. Instead, I learned that much of modern AI depends on fundamentals: mathematics, engineering, careful experimentation, and collaboration between researchers from different fields.

The biggest surprise was that AI research is not primarily about chasing intelligence. It is about solving problems. Researchers spend their time asking precise questions, collecting evidence, and testing assumptions. The process resembles science far more than the futuristic narratives often associated with AI.

So, what will Academia look like with the rise of LLM? It still seems to be the slowest to adapt to the various “hypes” and polished advertisements presented by the mainstream media, tech gurus, and big names.

The gap between AI hype and AI research is not a sign that the field is overestimated. Rather, it reflects the difference between public perception and the reality of scientific progress. The headlines show the destination; research reveals the journey.

Mathematics, programming, problem-solving, and critical thinking form the foundation of modern AI. Understanding how algorithms work, how data is processed, and why models behave the way they do is far more valuable than simply learning how to use the latest tool.

Research is full of experiments that do not produce the expected results. Progress often comes from understanding why something did not work rather than celebrating when it does. Curiosity and persistence are just as important as technical ability.

The future of AI will not be built by people who follow every trend. It will be built by those who understand the fundamentals deeply enough to create the next generation of ideas.

reDesigning the thoughts

Lazar Tomić — Fri, 19 Jun 2026 15:46:31 +0000

I spend most of my days inside a system I don’t fully trust.

It tells me what to learn, when to learn it, and how success is measured but it rarely explains why this version of learning is the correct one.

And the more I pay attention, the more one question keeps coming back: What if the system isn’t broken, but working exactly as designed?

I’m a student and a writer. Those two roles don’t always sit comfortably together. One asks for answers. The other keeps breaking them apart.

That tension became a project I call “Dumb by Design.” Not a complaint. Not a manifesto. More like a record of what happens when you start questioning the structure you’re supposed to succeed in.

Sometimes it’s critique. Sometimes it’s ideas for alternatives. Sometimes it’s just noticing things that feel off but are rarely said out loud.

I don’t know yet where this will lead. But I’m writing it here as it develops before it becomes something polished enough to stop feeling honest.

If you’ve ever looked at the way we learn and felt a quiet contradiction sitting underneath it all, you’ll probably understand why this exists. https://a.co/d/0jfqHFPa

#GET_INvolved

Lazar Tomić — Fri, 05 Jun 2026 19:00:59 +0000

Lazar Tomić

Jun 5

Inside My Internship at One of Europe’s Leading Physics Research Centers

#science #programming #productivity #python

5 min read

Inside My Internship at One of Europe’s Leading Physics Research Centers

Lazar Tomić — Fri, 05 Jun 2026 15:50:32 +0000

When one speaks of science, especially physics, the list of the names catching an eye always look like this: CERN, Lawrence Berkeley, GSI/FAIR, JINR Dubna…

The scientific output from such places has revolutionized the world, whether we speak of World Wide Web, Hadron Therapy, Cyclotron, Higgs Boson, anti-sigma-minus hyperon, or superheavy elements like Dubnium or Darmstadtium.

Here is how I, a high school student from Montenegro, associated my name with one of the most respected institution in science whose achievements shaped the world we know and became the youngest user of one of the most important scientific infrastructures in the world.

From 19th January to 30th March, I completed an internship at GSI Helmholtz Centre for Heavy Ion Research in Darmstadt through the prestigious GET_INvolved program, which is typically intended for master’s and doctoral students. It is a great honor for me, as a very rare case of a high school student, to have had the opportunity to become one of the users (youngest) of the institute’s scientific infrastructure through this program.

GSI holds the record for the discovery of the largest number of nuclear isomers (192), has discovered six superheavy chemical elements, and is currently constructing FAIR (Facility for Antiproton and Ion Research), a major international research infrastructure that is positioning the institute among the world’s leading scientific centers.

I worked under the supervision of Dr. Maksym Teklishyn and Dr. Valerii Panin on two different experiments, which is also relatively uncommon, as participants are typically assigned to a single project.

I contributed to both the CBM and NUSTAR collaborations. Through FAIR, four major experimental programs are being developed: CBM, NUSTAR, APPA, and PANDA.

CBM (Compressed Baryonic Matter) investigates a state of matter in which baryons (protons and neutrons) are compressed under extreme pressure and density to such an extent that their internal structures are modified or even broken apart. A simple analogy would be a sponge filled with many tiny air pockets. Under moderate pressure, the sponge compresses but returns to its original shape once the pressure is removed. However, if subjected to an enormous force far beyond everyday conditions, all of the air pockets would disappear and the material would become completely compacted. Its internal structure would change so dramatically that it would no longer retain its original properties.

Similarly, in compressed baryonic matter, protons and neutrons, which normally possess well-defined internal structures, are squeezed so intensely that their internal composition changes or may even dissolve into a different state of matter. Under such extreme conditions, matter no longer behaves as it does in ordinary environments but transitions into an entirely different phase.

NUSTAR pursues a different objective. It studies individual atomic nuclei, particularly rare and highly unstable isotopes that do not naturally exist on Earth. Through these investigations, scientists seek to understand how chemical elements are formed in stars and supernova explosions, how nuclei behave far from stability, and what the internal structure of atomic nuclei looks like under such conditions.

A simple analogy would be attempting to photograph a droplet of water at the exact moment it disperses into mist. The droplet exists for only a fraction of a second before disappearing. Scientists in NUSTAR do something similar on a much smaller scale: they study atomic nuclei that exist for extremely short periods of time in order to understand the origins of the elements and the fundamental structure of matter.

During its history, GSI also operated scientific infrastructure for cancer treatment using hadron therapy. Unlike conventional radiotherapy, which often damages surrounding healthy tissue, hadron therapy employs ion beams (such as carbon ions) that can target tumors with remarkable precision while minimizing damage to healthy cells.

Only four such centers currently operate in Europe, and only thirteen worldwide.

This treatment method is particularly effective for tumors located near sensitive organs and has demonstrated significantly improved outcomes. Five-year survival rates can reach 80–90 percent, and in some cases approach 100 percent among younger patients, compared with substantially lower rates achieved through conventional radiotherapy. Efficacy and safety of carbon ion radiotherapy for bone sarcomas: a systematic review and meta-analysis.

In addition to working on experiments, I participated in meetings of advanced scientific working groups within both the R3B Collaboration and the CBM Collaboration. These experiences allowed me to follow discussions on state-of-the-art technologies, recent scientific results, simulations, and comprehensive data analyses.

What was especially valuable was the opportunity to observe the broader scientific process: how researchers analyze data, formulate questions, think creatively, challenge assumptions, and exchange ideas. This perspective provided insight into how modern science is conducted beyond textbooks and classrooms.

The data-analysis tasks significantly improved my scientific computing skills, as well as my understanding of the physics and mathematics underlying the code. I noticed that I began approaching problems and searching for answers in a fundamentally different way. At the same time, I substantially strengthened my knowledge of the Python programming language and became more capable of connecting different components of complex systems while evaluating the strengths and limitations of specific computational approaches.

Working within a Linux environment and using modern computational tools was highly inspiring and further reinforced my decision to pursue a future in research and technology.

It is worth emphasizing that research differs profoundly from conventional learning, repetitive problem-solving, or even academic competitions, where solutions are already known and participants strive to find them. In research, only a small fraction of ideas can ultimately be transformed into meaningful results. Many concepts appear promising but fail to produce significant outcomes.

As science advances, an increasing number of questions seem already explored or partially understood, and researchers often spend years searching for the next genuinely new idea capable of changing the way people think about a particular field. For this reason, most experiments are, in a sense, destined to fail. Either resources are limited, or the experiment does not yield a substantial scientific contribution.

Research is therefore among the most demanding professions. The task of a researcher is not merely to perform experiments, but to continuously think, question, analyze, and learn. Scientific progress depends on the willingness to confront uncertainty, accept failure, and persist in the search for knowledge despite the possibility that the answer may never be found.

(Take a look at some of the interviews emphasizing my scientific journey!)
DOBAR DAN
Dobro jutro, Podgorice/ Evropski program za mlade istraživače pri GSI FAIR centru/ Lazar Tomić