Ich habe 17 Jahre lang für IBM gearbeitet. Die Einführung von KI erfolgte früh und "Null-Client" (Implementierung im eigenen Haus vor dem Verkauf an Kunden) begann fast, sobald Arvind Krishna die Geschäftsführung übernahm.
Die ersten Erfolge bestanden darin, Papierschieber wie HR und Finanzen zu ersetzen.
Diejenigen, die noch dort sind, soweit ich weiß, haben nicht festgestellt, dass der Ersatz von Programmierern durch KI zu Produktivitätssteigerungen führt. Qualität, Lesbarkeit und Konsistenz des Codes leiden darunter.
Augmentierung Erfahrene Programmierer profitieren beispielsweise davon, dass sie ihre Zeit für Dokumentation und Tests verkürzen und diese auf KI übertragen können.
Aber ohne die nächste Generation von Entwicklern können sie nicht skalieren, also stellen sie ein, um zu wachsen.
Und die GEN
IBM hiring triples. Has found the current limits of AI to replace workers.
byu/Superb_Raccoon inFuturology
8 Kommentare
Next generation of skilled programmers not looking too likely
Some of the best news us devs have heard in awhile. Here’s hoping the trend continues.
Anyone in the trenches of code knows that AI isn’t going to full-on replace devs anytime soon, augment? Sure. The thing is, most of the work isn’t even really code. It’s more like interfacing with stakeholders, colleagues, different teams, managers and clarifying requirements and processes. As much as devs get associated with code it’s a lot more coordination and acting as a liaison between what the client wants and what’s capable within the current environment.
I thought you said triplets, smh. I thought me and my two brothers were about to hit the big times.
I’m hoping that we’re in a phase of bad overly optimistic AI implementation, and then the economy will learn the lesson the hard way that that leads to bad results, then implement AI in a more cautious nuanced way that doesn’t cut out humans altogether.
Think of it as like the beginning of WWI where even experienced generals tried storming fortified trenches with wave attacks before wising up to the new reality of war. They mostly learned their lesson… albeit not before throwing a bunch of people in the meat grinder.
I guess I’m saying that the current situation sucks, but it doesn’t have to be a linear path all the way to Hell. From this post, it sounds like IBM is in the phase where they try to do better.
Yes. For all the doomsayers out there, what they don’t realize is AI will allow for workers now to focus on other things making them more productive. We’ve already been through several of these situations in the past. The “green revolution” in agriculture fundamentally changed the workplace in the early to mid 1900’s from an agricultural based economy to an industrial economy. Once people could get off the farm then more engineers, scientists, etc happened and look how fast technology moved once that happened.
Combine the fact that we are dangerously close to negative population growth (which means far fewer younger people) and AI is not a threat. It’s a necessity.
AI will replace programmers the same way excel replaced accountants. It makes a lot of entry level stuff reachable for the average person, and is a huge force multiplier for those who know how to utilize it. Those who fully ditch their devs will eventually suffer with compounded technical debt. But eventually, being able to use the tool properly will become a requirement to be effective in the field.
Idk, this is hard to say. There seems to be still a lot of room for improvement in test time side of things.
We hitted the wall regarding LLM architecture but there are infinite number or possibilities with test time compute.
So i’m sure it’s possible to compress that test time logic into some new model architecture that merges llm layers and ”deep thinking” layers… so it will be even faster and easier to do more test time compute.
So we really don’t know what the endgame is
Anyone referring to other humans as paper pushers deserves to be replaced by a computer they built