Member Avatar for Ying_Yang

I'm thinking of something to create to improve computer's classic software, maybe a higher level programming language than the actual, or a smarter OS, or maybe make expert systems improve with AI, etc ...

On the other hand, Computers nowadays can be very helpful, but also can be used to steel information, terrorism, and other crimes.
If we do make better computers, they will eventually fall in the wrong hands, and be a more efficient tool for bad.

How do you think we can prevent it ?

I believe there might be a way to prevent it, but not by programming security policies, because that can be easily get passed through back doors.

I believe there must be a secure way to make better computers :)
What ideas do you guys have ?

  • 4 Contributors
  • 6 Replies
  • 235 Views
  • 1 Week Discussion Span
  • Latest Post Latest Post by Ying_Yang

Dani AI

Generated

As opened the thread, there are really two goals running side‑by‑side: make machines that capture human intent more reliably, and make every advance harder to turn into a weapon. ’s point about context-sensitivity is exactly the right place to start — the technical work is about encoding intent, and the security work is about constraining where and how that intent can be executed.

A practical technical path:

  • Push “intent-first” design: build small domain‑specific languages (DSLs) or strong type/contract layers that let programmers describe what they want, not just how to do it. That reduces accidental ambiguity and makes automated checking feasible.
  • Treat critical modules as verifiable artifacts: property tests, model checking, or formal proofs for the smallest, highest‑risk pieces. Keep the verified surface small and well specified.
  • Prefer memory-safe languages and effect/type systems that express permissions and side effects. That moves many common exploits from runtime surprises into compile‑time errors.

A practical deployment and safety pipeline:

  1. Develop in a constrained DSL + strong tests.
  2. Produce reproducible builds and sign artifacts.
  3. Run modules inside a minimal trusted runtime (microkernel or compartmentalized runtime) with capability-based permissions.
  4. Use hardware roots of trust and attestation at boot.
  5. Enforce runtime quotas, rate limits and telemetry for anomaly detection.
  6. Keep audit logs and reproducible provenance for any sensitive action.

No single technology will stop misuse. Social controls matter: governance, access policies, red‑teaming, and transparent audits are part of the design. ’s call for smarter OS and extra checks maps directly into minimizing the trusted computing base and baking checks into compile/deploy pipelines. And as reminded us, dual‑use is nothing new — design choices and institutions determine whether an advance helps or harms.

Recommended Answers

All 6 Replies

Member Avatar for ddanbe

Did you know that in English there are about 30 different meanings of the affirmative word yes?
A story to illustrate:
A man comes home from his work, says hello to the wife, installs himself in a seat with a newspaper and a drink.
Two options:
The wife says: " Darling, are you going to mow the lawn? "
The wife says: " Darling, I'm taking a bath will you come and rub my back? "
In both cases the man will say " yes ".
The moment we have a computer who understands the difference between the two affirmations, we have something that is close to an ideal computer.

Member Avatar for Ying_Yang

Did you know that in English there are about 30 different meanings of the affirmative word yes?
A story to illustrate:
A man comes home from his work, says hello to the wife, installs himself in a seat with a newspaper and a drink.
Two options:
The wife says: " Darling, are you going to mow the lawn? "
The wife says: " Darling, I'm taking a bath will you come and rub my back? "
In both cases the man will say " yes ".
The moment we have a computer who understands the difference between the two affirmations, we have something that is close to an ideal computer.

Yeah, indeed that's one way to really improve computers, that field its not been studied much, that's why I'm thinking maybe that can be done if the programming languages could understand the programmers like that... then we can have much smarter OS.

Member Avatar for albertjone125

The designing of smart OS systems and more security check while making a program on specific program may be considered then it will relieved to some extent.

Member Avatar for Ying_Yang

The designing of smart OS systems and more security check while making a program on specific program may be considered then it will relieved to some extent.

I'd apreciate if you could explain that in more detail, I didnt understand.

Member Avatar for jbennet

Computers nowadays can be very helpful, but also can be used to steel information, terrorism, and other crimes.
If we do make better computers, they will eventually fall in the wrong hands, and be a more efficient tool for bad

Well, the whole reason computers exist as we know them, is because of the second world war (namely Colossus and ENIAC).

Member Avatar for Ying_Yang

Well, the whole reason computers exist as we know them, is because of the second world war (namely Colossus and ENIAC).

But I believe we can take it to a better end :)

Be a part of the DaniWeb community

We're a friendly, industry-focused community of developers, IT pros, digital marketers, and technology enthusiasts meeting, networking, learning, and sharing knowledge.