opinion Congratulations! As a CEO of a huge tech company, head of a sovereign wealth fund, or a VC tired of Megayahat's jumping, you can spend billions of other people's cash. You want to make a difference. You want to succeed.
Above all, you want to do it and look better. Currently, nothing looks better than AI to do all three.
This columnist understands your needs. Nothing is a future-oriented builder than a big splash of artificial intelligence. It is an exciting field in an exciting time, and promises a major cash return and a stronger reputation for large investors. But I would fail my duty, but we had to provide guidance to do a fair deal. It means understanding the pitfalls.
They can't find that they've swept $5 billion into the blank and bought it to a bankrupt ordure abandoned man. You may not care much if you consider your investment primarily as marketing and carefully evaporate your money into the pockets of advisors, partners and consultants with all workers years later. In that case, you know your business better than we do. enjoy.
Builder.ai coded itself into horns – now it's bankrupt
read more
However, if you're piloting a huge tech company with all-in AI, investing in AI obstacles can look bad, very bad and really bad. Is it bad that you don't understand how to evaluate AI propositions, or is it not AI able to live up to the hype? If AI investment was in your specialized Bullseye code generation technology, what's really awful is that just a few years ago you decided to coordinate your core mission and integrate high-tech with flagship cloud services and AI platforms. Seriously, how bad is that? Yeesh.
So, if you don't want to align your core mission to bankruptcy, here's what to do. First, we will perform appropriate analysis of the technology. Is it the real thing or is it a branding layer above offshore humans? This is a bad sign. If it appears to be trying to do what it claims, is it working? Are you even sure what this means here? Yes, this is difficult with AI. This is especially difficult if you don't have something that works all the time or generate no-code AI apps. There's no such thing as an approach to code generation that always works, so what are you buying?
Like large tech investments, wise organizations argue proof of concept pilots before looting small cash funds. POCs are forensic design and can be meaningless and misleading in themselves if the results are not rigorously verified. As the more keen observers have already pointed out, forensic rigor is not a major attribute of business AI. This explains a lot.
As an example, you can quickly return to the Turing test. This is a loosely transformed thought experiment, forced into an eternally unpleasant construction to prove something. When AI is most successful, it is again widely pointed out, and is strictly designed to perform specific tasks with quantifiable outcomes. Code that generates AI certainly generates code, but that's not the point. For example, 30% of the codebase is generated by AI, meaning nothing without context. Windows before the windows was mostly written in the assembler and was a very bad idea to force a complete redesign.
The code that generates the AI is not a standalone black box app generator, but a team member who needs expert human collaborators. If that speeds up their work and quality of UPS, then it's fine. If not, it is a problem and not a solution. If you are coding in an organization that is politically impossible for your artificial helpmate to report anything, both you and your senior management have very real issues.
The only useful POC for AI coding is where the time and quality of generating apps and their quality actually did the job compared to the total cost of a carbon and silicon mix. This means a standard app that can be built by combining methods in public places where nothing can be hidden and competition is all about. The cybersecurity team won the flag, and the self-driving cars were born at the DARPA Grand Challenge. Generating AI code requires a unique robot war.
That's the last hint for those who want to make a difference. It's about success and looking good with AI. Before you drop a bundle into a marketing-driven hype machine with the inherent health and lifespan of chocolate cakes in kindergarten, spend just a small portion of it in a big public competition where app teams can convene with the best AI coding help they can convene. This proves that everyone in POC, especially you, need to validate AI-assisted coding and create a framework for future development. That is, of course, if the outcome of such an idea can't make you even worse.
You know your work better than we do. ®
