The most expensive AI decision may be the one nobody decided to make.

A change can be real before its authorship is clear

A model recommends a refactor. Someone accepts it. A test passes. The next release moves forward. Did a person choose the behavior, did the model, or did a sequence of routine steps produce an outcome no one intended? That question does not have one answer across all projects. The point is that the system and the people who depend on it can be affected while the answer remains uncertain.

AI can also make a team more capable. It can reveal an alternative, bring an old codebase into view, make a prototype possible, or help a skeptical reviewer ask a better question. AI4DEV is a live seminar about using AI for software development and learning to recognize where its help is valuable, what evidence matters, and what human work surrounds the result.

Follow the consequence

A coding shortcut becomes a review habit. A demo becomes a product expectation. An AI-generated test passes and everyone feels a little more certain. In each case, the people involved may have acted reasonably. Yet the next release can still contain a choice no one remembers making—or a result that seems to have chosen a direction for them. What, exactly, changed between the first suggestion and the commitment?

The question reaches from software work into customers, data boundaries, adoption disagreements, careers, and leadership duty. AI4DEV gives professionals a stronger way to examine the work while the next answer can still be questioned. Developers can ask what the code does. Managers can ask how a divided team learns. Product colleagues can ask what the prototype proved. Leaders can ask what is already changing. Each question gains from the others.

Where did the choice enter the work?

The guide has built through more than one AI era

Howard's record reaches from LISP and FORTH, MIT supercomputer research, and NASA Ames parallel computing through commercial products and enterprise systems into current AI-assisted engineering. In 51 paid years and more than 200 engagements, principally through companies he founded, he has helped clients move new technology into work other people had to depend on. GM and Ford needed complex factory designs made visible so their executives could approve; Howard built the animated virtual reality web signoff system. His record also includes repeated advanced-architecture teaching for Valtech and recent AI-assisted prototypes for Telos Alliance.

The Telos AI-for-product-development seminar expected about 20 participants and drew 47 across development, product, management, executive, and skeptical roles. Client president John Schur credited Howard with meeting those perspectives and wrote, “I felt that Telos received multiple times the value of our investment.” What might a room with your own builders, skeptics, managers, and product colleagues be able to see together?

Explore the work behind Howard's perspective. · Read John Schur's full recommendation.