Hacker News Reader

Comment by gozur88 | Hacker News Reader

>No amount or greater sophistication of the algorithmic structures listed above, can replace genuine human nuance, interaction and trust.

I don't think this is true. Not at all.

View on HN

I think there is a technical and experiential aspect to these 3 elements. I think the technical aspect can be met, but perhaps it's equally true that the experience of trust etc with another human can not be replaced (in a literal sense, given that it specifically requires a human).

Me neither. How much of the genuine human nuance is present in today's financial services or any commission driven industry - next to nothing. Every agent/broker is motivated by the highest commission he/she can make, nothing more.

Even if this was true (which it isn't: in most areas of finance blindly following this strategy is a rapid route to getting sued for breach of fiduciary duty) there's a lot of human nuance involved in actually selling the service to a client who won't take "well the model isn't tractable but here's last year's results and an article on ML" as an answer

So what is a more reasonable answer that the client will accept ? Ultimately, it's a pattern of words and data. Just playing the Devil's advocate here. Any ML system that can be trained on such patterns should be able to learn the nuance, per client. No ?

I'm certainly not convinced any ML system can pass a high-stakes, niche-interests version of the Turing test (and since winning people's trust back after losing their money is an emotionally charged thing, probably something akin to Philip K Dick's Voigt Kampf empathy test too). Even if they could, you haven't got a training set because (i) client conversations aren't usually on the record (ii) none of the previous human conversations that could be used if they had been recorded were remotely related to the whys and wherefores of the investment decisions the algorithm actually took (iii) the investment model generated by the ML process probably isn't tractable enough for even its own human designer to convert into words what it's actual strategy was and will be in the next period.

You don't need "any ML system" to be able to not only process the data, but also explain the strategy behind its data processing at various levels of granularity, answer abstract questions and do a convincing impression of listening and responding to client feedback, you need advanced general intelligence.

As for what the more reasonable answer a client would accept is, I don't think I really have the requisite years of investment banking experience to know that...

I can reasonably say that you make this comment because you know nothing about the business logic of trading desks.

A trader is not necessarily a sales or a broker, though. Nor is he a quant, or a dev. People rarely imagine how many different jobs are involved in the trading job, and how rich the business logic is. At my shop, traders are the piece that connects all of the jobs in the value chain.

I believe, currently, the business logic can be improved locally by learning systems (and it is), but there is no public example of an industrial learning application encompassing a scope comparable to what the usual trading desk handles. Sure, there are many inefficiencies ; traders work on heuristics, afterall. But I don't believe we have the necessary horizon to aptly predict the end of traders, because I don't see how we could make AIs with a better efficiency.

I'd say I know something about the business logic of trading desks. Some of my friends worked at GS. I'm not implying traders are brokers. I do agree that the trading business logic is very rich and that AI will be nowhere close to replacing a trader in the next 15-20 years at least.

I do foresee a future in which a trader can accomplish a lot more than he/she can today using AI. So in the future as the per trader efficiency increases, the number of traders required will most probably decline, unless there is a dramatic increase in trading volume that cannot be matched by the then state of the art AI.

This is essentially what's happening today with X.ai, Facebook messenger and the like. Sure the logic involved in booking air tickets for a group of 5 over 10 conversations isn't as complicated as the rich business logic of a trade, but 5 years back Facebook messenger would've seemed almost impossible, just like the trading business logic seems impossible to do with AI today.

I have no doubt new technology will be leveraged to improve productivity, but that's been quite common in the last decades.

On the other hand, the technology advances needed to transform the tools into standalone actors are not merely a matter of scaling current technology. Especially, the creation of training datasets is a problem for which we currently have no solution, that's why we fall back on human trainers (mturk, etc). That's why a solution to a problem with no clear, bounded model and no easy dataset seems out of reach.

Reply on news.ycombinator.com