minware reposted this
2024 was the year of reckoning for AI. My most-read posts by far (50k and 20k impressions for #1/#2 vs. 5k for #3) were about how AI is only as good as a very junior engineer, and how we’d be better off automating easier tasks (like labeling Jira tickets) than hard ones (like building software). My friends and family range from total disdain (“I hate the tab suggestions in gmail — I know what I want to write!”) to extreme optimism (“AI can’t replace software engineers today, but just wait until next year!”) LLMs are unquestionably the most powerful advancement in computer science in the past decade, but are clouded by hype. People don’t understand how to use them yet, and many billions of venture capital are flushing down the drain in a wave of FOMO and euphoria. In 2025, winners will emerge who use this new tool well, and those running on hype capital with vague promises of replacing engineers will be seen for what they are. Which side are you on?
There is a lot to think about here. Even a required feature in nearly all software (authentication) isn't something with a "standard" implementation we can depend on LLMs to produce. Passwords or SSO? Self-managed with passwords + OTPs or managed SSO? Social or enterprise SSO? Passkeys or magic links? Are the LLMs immune to advertising? If I prompt for some SSO code, will I always get a managed solution over OSS? Will the "suggested" managed solution be the highest bidder, per the usual rules of the game? By the time I optimize the prompt to remove the noise, I've done a good chunk of the decision-making already.
Great perspective, Kevin! I agree that while LLMs are powerful, there’s a lot of hype that hides what they can realistically do. It's important to focus on using them for easier tasks instead of expecting them to replace complex engineering. I’m curious, how do you think smaller teams or startups can use AI wisely without getting lost in all the excitement and promises?
I feel like we're not quite to the "there's no bad ideas" phase of the hype (which fostered the dot bomb), and LLMs are fun to use and/but are expensive. Expensive to train and expensive to execute. My guess is we've pretty much plateaued on what the technology can do in its current iteration, but there are still a ton of specific applications, efficiencies, and focused stepwise improvements that will come through research. I think what will be interesting to see if we start to see improvements in efficiency that rivals Moore's law or if we see a different trajectory. I think the most successful company in this space will lean into efficiency to boost margin and continue to make it easier to specialize and integrate their *LM offerings with a flexible but robust free/commercial licensing scheme. Following close behind are analytics supported answer models like Bast AI and a cross between expert systems that have dominated medicine and offerings like Bast.
Completely agree that a lot of AI’s potential is being overshadowed by hype right now. It feels like everyone is chasing the 'next big thing' without pausing to figure out practical, sustainable applications.
I'm glad somebody's saying it. It can be a useful tool in an engineer's hands, but it's for augmentation not replacement.
Hello IHello!) I'd like to work at your company remote.))
I don't get out of bed in the morning until I ask AI what I should have for breakfast
Turning Small Businesss into Multi-Million $ Businesses | Proven Strategies | ActionCOACH
2moIn 2025, Kevin Borders, what strategies will differentiate successful AI adopters from those riding on hype?