menu

Help me build a map to navigate generative AI

BackerLeader ●15 ●110 ●208
calendar_today • schedule1 min read

Back in April I shared how I'm grappling with Generative AI. I wanted to get away from the hate-or-hype cyclone of doom and ask some questions that would be useful to navigating this technology.

  1. What harms do we see now, or predict in the future, that we could avoid?
  2. Who is responsible for each of these harms and could they take action now rather than in response to controversy and backlash?
  3. If there were no harms, what use could we make of this technology and it is fit for such purposes?

I'd love to hear your thoughts on these 3 questions!

Here's an example answer... I'd love to hear your thoughts.

The training data includes copyrighted material used without permission (1) and the people selling models should pay to license this data (2).

I would use AI to help me code, but never to help me write an article.

4 Comments

•

reat perspective. Generative AI is evolving so fast that having a clear map of tools, use cases, and limitations is often more valuable than mastering any single platform. The key skill is learning how to learn and adapt as the landscape changes.

1 vote
•

A practical way to navigate generative AI is to focus on the fundamentals first—LLMs, prompting, embeddings, agents, and workflows. Once you understand how these pieces connect, the AI landscape becomes much easier to explore and evaluate

0 votes
•

Good framing — especially the separation between "what harms exist" and "who acts on them."
On question 1: the harm I see most in our domain (Web3/security tooling) is false confidence. AI-generated risk scores that look authoritative but aren't auditable. Users trust the output without understanding the reasoning. That's more dangerous than a wrong answer — it's a wrong answer that feels right.
On question 2: responsibility sits with the builders. If your model produces a score, you owe the user an explanation. We made interpretability a core requirement early — every risk assessment shows exactly which signals triggered it and why. Not because regulators demanded it, but because a black box in a security context is a liability.
On question 3: AI for pattern recognition at scale — yes. AI for final judgment — no. We use ML to surface candidates for review, not to replace the review itself.
The map you're looking for might just be: the closer the decision is to irreversible consequences, the more human oversight it needs.

1 vote
•

@[VeritasLab] thank you, you've provided lots of great thoughts for me to incorporate. You've come up with some unique things I haven't thought of!

1
🔥 Join developers growing publicly
Share your knowledge, build in public, and grow your developer presence with a global community.
Join CoderLegion

More Posts

The Sovereign Vault — A Comprehensive Guide to Protocol-Driven AI

Ken W. Algerverified - Jun 4

I’m a Senior Dev and I’ve Forgotten How to Think Without a Prompt

Karol Modelskiverified - Mar 19

AI Reliability Gap: Why Large Language Models are not for Safety-Critical Systems

praneeth - Mar 31

Your AI Doesn't Just Write Tests. It Runs Them Too.

Kevin Martinez - May 12

What Is SARIF and How Does It Help Security Tools Work Together?

Ganesh Kumar - Jul 4
chevron_left
Steve Fentonverified
14.2k Points • 333 Badges
UK • stevefenton.co.uk
94Posts
145Comments
67Connections
Steve Fenton is an Octonaut at Octopus Deploy. He’s a Software Punk, author, programming-architect, ... Show more

More From Steve Fentonverified

Haute Culture!

Modernizing Legacy Software

The Compliance Ratchet

Related Jobs

View all jobs →

Commenters (This Week)

4 comments
2 comments
1 comment

Contribute meaningful comments to climb the leaderboard and earn badges!