Another way of saying it, if an AI agent doesn't have true risk of oblivion (or mortality in the biological sense), it will not be incentivized to avoid it (or develop the ancillary processes associated with this avoidance ie desire to self-replicate).
In my own experience, nano banana still has the tendency to:
- make massive, seemingly random edits to images
- adjust image scale
- make very fine grained but pervasive detail changes obvious in an image diff
For instance, I have found that nano-banana will sporadically add a (convincing) fireplace to a room or new garage behind a house. This happens even with explicit "ALL CAPS" instructions not to do so. This happens sporadically, even when the temperature is set to zero, and makes it impossible to build a reliable app.
Higharc | Senior Frontend Engineer, Senior SDET | Full-time | Durham, NC, San Francisco, CA, remote | higharc.com
Higharc has built a cloud platform for homebuilders (build >80% of homes built in the US every year) that massively reduces the burden of producing construction documents, material counts, and sales tools. At the center of the platform is an in-browser CAD system that allows builders to model homes with the information needed to support the complexity of their business. We raised a $53 million series B in 2023.
We’re building a complete home building web platform. Our system, built with TypeScript, C++, and WebGL, produces permit-ready plan sets, bills of materials, and beautiful, realtime product configurators with realtime estimates/BOMs. We just raised a $21m Series A and are hiring across a wide range of engineering roles and seniorities.
We are a remote-first team (w/ offices in Durham, NC, SF, and Atlanta).
Roles:
- Frontend engineers (next.js, React, pure TypeScript)