Most of the biggest AI failures of 2024, as cited by MIT Technology Review, are related to our collective inability to turn AI into applications that regular users want. Even something as basic as Apple’s attempt to summarize incoming text messages is more confusing than useful. Some of this is due to a lack of trust that AI will deliver consistent results, but much of it is a result of constantly burdening developers with AI fundamentals. It forces them to do the heavy lifting of turning the fundamental building blocks of AI, similar to cloud primitives like storage, networking, and compute, into applications.
What AWS and Microsoft Should Do
With so many AI options, you may need AWS or Microsoft. I say “probably Microsoft” because it seems like the market needs something similar to what Microsoft did for networking beginners: clear documentation, intuitive user interfaces, etc. AWS achieved great success in the first decade of cloud computing by providing developers with the same fundamentals they were used to – the same LAMP building blocks they used on-premises – while adding the flexibility of Elastic.
For contrast, read Amazon SageMaker’s marketing description. AWS talks about a “unified environment for analytics and AI with unified access to all your data,” using “familiar AWS tools for model development, generative AI, data processing, and SQL analytics.” That’s a good thing, but it doesn’t force developers to learn new tools.) But AWS falls into the trap of claiming that developers want and need “purpose-built tools.” “Built for purpose” can feel like a euphemism for “give it all,” so deciding which model to use can feel like a coin toss rather than a clear decision.
Again, Microsoft has achieved great success in the areas of networking, operating systems, and developer tools by providing easy-to-use options for mainstream IT managers, developers, and others. It never really appealed to tech industry leaders Alpha Geek, but why did it succeed? The real money doesn’t come from appeasing AlphaGeek’s greed for arcane options with infinite configurations. Because the real money comes from providing an easy option for people who love technology, but who are more important to getting home from work on time to spend time with their families. Please keep in mind that it’s AlphaGeek talking about the latest and greatest models from X (or Mastodon or Blue Sky). It’s a good idea to provide them with rich AI raw tools. However, for mainstream developers and data scientists, the real opportunity will be if cloud service providers do not force AI application development.
dl-itworldkorea@foundryco.com
Source: www.itworld.co.kr
