Also relevant, as far as being just as woke and yet not going broke:
The newest Claude is apparently super impressive for programming, to where programmers who aren’t also architects should wonder if they’ll be needed in a year or two.
1 Like
I do not know the details of the agreement between the government and OpenAI, but secretary of war Hegseth stated that the government’s objection to the Anthropic agreement is how it is enforced. He stated that Anthropic wanted to retain a veto over how the technology is used. The government has stated that they will use it lawfully. Apparently this was enough for open AI so they came to an agreement.
There’s probably one architect for 50-100 programmers. But also LLM is at a level where it can replace low level architects. I played with Google AI Studio recently to build a simple app. I just described what I wanted and it spat out a functional product. It basically loops through a design-implement-verify-test cycle until it finds the answer satisfies the request. Kinda scary for us software people. The only saving grace at this time is that due to the small-ish context window, it’s not able to maintain very complex software with a huge source base. But that’s probably coming soon enough.
1 Like
I have a hard time thinking how this is different from any regular user license agreement for other software. These agreements are basically forcing the user to comply with uses allowed by the software provider or stop using the software. Maybe the distinction is that Anthropic wanted to be able to validate each use beforehand, where as Open AI reserves their right to stop the government from using their LLM after they are proven to have used their AI in ways not agreed upon in the contract. Still seems like a marginal difference thinking about what OpenAI and Anthropic stated their identical red lines were.
I’m not a programmer but I used Claude CoWork recently and it took no time to build me an app with nice responsive GUI to do a relatively simple task (portfolio updating pulling daily quotes, and rebalancing across taxable/tax-deferred/tax-free accounts to optimize asset allocation and location, with options for ETFs that would allow such rebalancing, along with several graphical reports). Took me way longer that I’m sure it’d take a programmer but I didn’t have to write a single line of code, just iterate until I got the functionality I wanted. Probably won’t be able to build my own Empower or Quicken replacement just yet - mostly because I don’t have the time or drive to do it - but if I were these software providers looking at what a total noob can do in a day of work, I’d be a bit concerned about my future monetization opportunities.
2 Likes
Some providers are doomed. I was thinking someone could replace TurboTax real quick to offer the same service for MUCH less. It’s a huge project, but it’s probably quick and easy to implement with LLM assistance.
I’m concerned about MY future monetization opportunities
Transformation in many knowledge/service related areas could be as much as you’d see in a steel/aluminum foundry from 50+ years ago to now. From 100s of workers manually enabling the production to a handful of engineers at their consoles controlling the whole production remotely 24/7. Definitely uncertain future for my own line of work (although I’m FI and about to retire so …).
1 Like
The current transformation seems significantly faster, quite possibly faster than we can peacefully adapt to it.
No question it’s significantly faster. I’m curious though about what non-peaceful adaptation would look like besides falling irreparably behind the countries that do manage a peaceful transition. Maybe deploying autonomous weapons to implement long-lasting entitlement reform…
A little more detail on the opportunistic deal from OpenAI: updated terms
Honestly, I don’t find the language very re-assuring:
The AI System will not be used to independently direct autonomous weapons in any case where law, regulation, or Department policy requires human control, nor will it be used to assume other high-stakes decisions that require approval by a human decisionmaker under the same authorities. Per DoD Directive 3000.09 (dtd 25 January 2023), any use of AI in autonomous and semi-autonomous systems must undergo rigorous verification, validation, and testing to ensure they perform as intended in realistic environments before deployment.
So once fully tested and validated, then it’ll be ok? Feeling safer already knowing that.
The AI System shall not be used for unconstrained monitoring of U.S. persons’ private information as consistent with these authorities. The system shall also not be used for domestic law-enforcement activities except as permitted by the Posse Comitatus Act and other applicable law.
Considering recent deployment of the National Guard in violation of the Posse Comitatus Act, I’m sure nothing nefarious will come of it…
I was thinking along the lines of high unemployment → civic unrest. The fuse is already very short, now imagine if another 5-10% of the population didn’t have a job.
2 Likes
Can you explain why you need this and how you’re using it? Pulling quotes is easy in Google Sheets, there’s a googlefinance function for it. How did you do the rebalancing part? I’m guessing you didn’t let it actually rebalance your account but rather make suggestions? How did you get it to find compatible ETF options?
I’m using it to get suggestions for how to rebalance my portfolio to a specific mix of equities and bonds, but also within equities, specific targets for international, small cap, and large cap value. One issue is that I want our tax-deferred accounts to hold all our bond allocation, but since my options there are limited by my employer, it makes balancing the rest of my portfolio a bit trickier.
The other part of our allocation target is to manage basis vs. capital gain in our taxable accounts (planning to withdraw from these while keeping AGI low enough for about 10-years of ACA credits milking). Another limitation was that we wanted to have it all done at 1-2 brokers. So I entered all these limitations and basis in various taxable accounts. I also specified that I prefer rebalancing using my Roth IRAs as much as possible, then tax-deferred, and finally taxable accounts. Maybe I should have an option to change the order…
As far as ETF options, I asked Claude to pull ETF options from Fidelity, Blackrock, Schwab and Vanguard limiting them to low-cost options and list the asset allocation of each. It should be easy to add more (especially if I wanted more exotic ETF like covered call or crypto ones) but that was usually sufficient to get a decent number of suggestions. App pulls these from websites at each broker, and pulls the asset allocation from FinanceCharts and a couple more sources. I spent time double-checking these on morningstar because I wanted to have reliable data for it and that turned out to be accurate enough for my purposes. Although this doesn’t change too much too often, I didn’t want to have it redo this every time the app runs so I just asked to make a button to refresh the asset allocation data on demand.
I also wanted an option to change my target allocation so there’s an input panel for these with option to add categories say crypto, gold, REITs, TIPS, etc…
After that, you’re right that all I wanted was to display the information for current asset location, allocation, and % departure from target. Above a threshold % departure from target (which can be changed for each asset class), app displays ETFs options that would fit and what amount to trade to get things back on target. I did not want to allow it to rebalance automatically for multiple reasons, not the least of which was security.
1 Like