Katrina Delos Reyes

Observatory

Observatory · Planned

Software Is Changing First

A longitudinal case study of AI in software engineering

By

Future of software engineering · AI and work · Coding agents · Engineering organizations

Domain
Software engineering
Status
Planned
Signals tracked
Capability · Behavior · Work · Economics · Institutions · Humans
Methodology
Public sources only. Each signal will list its source, date, and a stated confidence level.

Software engineering is the Observatory’s first domain because it is where AI-produced work is most visible today and where the effects can be measured in public evidence. Examples include changes in open-source activity, published developer surveys, hiring and job-posting data, and documented changes in engineering practice.

This case study is in its planning stage. The signals below are what it will track. Each one stays empty until there is public evidence to fill it.

Capability

What can AI systems do in real software work, as opposed to on benchmarks?

Capability · Real-world task performance

No data yet. This signal will be added when there is public evidence to support it.

Behavior

How are engineers changing how they write, review, and test code?

Behavior · Adoption and usage patterns

No data yet. This signal will be added when there is public evidence to support it.

Work

How are tasks, roles, and team structures changing? What is happening to early-career paths?

Work · Roles, tasks, and entry-level pathways

No data yet. This signal will be added when there is public evidence to support it.

Economics

What is getting cheaper, and what is becoming scarce?

Economics · Cost of producing versus verifying software

No data yet. This signal will be added when there is public evidence to support it.

Institutions

How are education, standards bodies, and regulators responding to AI-generated software?

Institutions · Standards, education, and policy

No data yet. This signal will be added when there is public evidence to support it.

Humans

What is happening to craft, identity, and responsibility among people who build software?

Humans · Qualitative accounts

No data yet. This signal will be added when there is public evidence to support it.

Written by , part of The Inner Intelligence Era.

Continue the investigation

  1. Research · Researching · Software Is Changing First

    We May Be Measuring AI-Generated Software Completely Wrong

    AI increasingly takes part on both sides of software development, generating systems and helping verify what was generated. What counts as trustworthy software assurance when implementation, tests, review, and evidence can all be produced by machines?

    AI-generated software · Software verification · Software assurance · AI code review