Dear Friends,
A couple of days ago, I was
walking down a lane when a group of five students was walking in front of me.
The discussion among these boys was about learning programming. It looked like
they were all first-year engineering students, presumably most of them from
CSE.
One boy was sharing his
experience of learning Python programming, for which he was prompting ChatGPT
for every programming assignment, whether in the lab or at home. Others were
asking, “What would you do in the main exam?” The casual response was, “Let's
see when it comes.” Another asked, “What would we do during the campus
selection process, as they expect DSA-level programming skills?” Another
casually answered, “That is a far easier process than a semester exam.” The boy
continued, “Most recruitment tests include online assessments, which we can
manage easily.”
Fever is not a disease. It is
a symptom of an actual disease. If we don't identify and address it at the
root level, we are going to lose a generation.
Most companies are aware that
cheating is happening through multiple means—LLM-assisted answers on a second
device, AnyDesk, remote software logins by another candidate, prompting, hidden
second devices (especially foldable devices), browser manipulations, hidden
notes and physical materials behind monitors or on walls, screen mirroring, and
more unwanted innovations.
Companies are ignoring this fact
because they mostly depend on second- and third-level physical interviews.
Having said that, what happens to the genuine student who has worked hard? Is
it not a great loss to organizations?
A few days ago, there was an
article in The Times of India (citing World Bank research) on learning
paralysis. Students are using LLMs, boosting their homework scores by 18% while
spending less time on assignments. The same students saw their in-class exam
scores drop by 20%, with overall entrance exam performance falling by 5–7%.
Researchers from UC Berkeley,
CMU, and MIT studied 1,222 people and found that using AI for just 10 minutes
made it harder to focus and solve problems independently. When the AI tool was
taken away, participants quickly gave up and made more mistakes on math and
reading tasks. The study suggests that relying on AI for quick answers can
weaken our own thinking skills over time. This is called Cognitive Paralysis.
What is the remedy?
Microsoft is working on it with Carnegie
Mellon University through a programme called PRAISE. It is focused
on accelerating junior engineers' learning in the AI era so that they build
real skills and don't just ship code faster. The focus is on the parts of the
job where people still matter most: understanding the problem before writing
any code, questioning what AI gives you, and learning the fundamentals that
keep you good at this for years.
Essentially, it boils down to “RAW”
skills—Reading, Arithmetic, and Writing. In computer science terms,
understand the problem, develop a flowchart, prompt it, and generate the code
using AI (productivity is the only merit factor).
The worrying part is that AI is
learning faster than humans. The layoffs have surfaced at the humanoid level as
well.
When Figure AI replaced its
Figure 02 humanoid robots on BMW's production line, it needed to retire the old
fleet without leaking valuable trade secrets. Following a viral suggestion on X
from Arnold Schwarzenegger, the company sent the fleet to a foundry in Finland
to be melted down. The robots were trained to leap into a 75-ton furnace of
molten steel, capturing the event on film with Schwarzenegger delivering his
iconic “Hasta la vista” line. They learned to jump and were melted down.
What is my general suggestion
to freshers?
Working for a startup gives them
a better experience than working for an established organization.
Reasons—opportunities to delve deep into problems from the early stages,
working with teams despite lower salary packages, opportunities to work through
a complete cycle (from problem identification to final deployment), and
sometimes, a disgraceful exit as well.
This is real learning.
This may not be possible for everyone in mature companies.
What is your current learning
experience?
No comments:
Post a Comment