Hacker News Reader

Comment by babyshake | Hacker News Reader

> 1. Sales -> You can hire a lot of sales people if they're generating enough business to cover their own salaries

It won't be long before GPT-3 or its successors will be putting many people out of a sales or support job.

Funny to see how little value devs and engineers put on sales people. Enterprise software success depends highly on sales people. Startup success depends highly on marketing. At previous company, SAP flew the CEO and a few others to Monaco for the F1 race. Interesting to see we had SAP and PeopleSoft implemented at the company.

Yup. Devs like to think that sales people don't contribute real value and are basically just marketing, and that's true to a degree, but it's also true that sometimes the entire product doesn't contribute real value either. A senior dev building Uber/delivery clone #1745 run by VC funny money is "contributing" the same as a sales person, no matter how many frameworks and algorithms he uses. (Unless, of course, that algorithm is itself contributing to some efficiency in the deliveries.)

So the value the sales people provided was successfully bribing the executives.

You're not exactly wrong here, but many that's a kinda hilarious example. AI doesn't yet know how to operate in the grey area of not-quite-illegal bribery.

Why can't a computer program automatically buy tickets for F1 and send them to prospects?

How will it determine who gets the tickets? And which event is the most appropriate one for a given prospect? How will it know that the prospect is interested in the first place? Or available? Or worth wooing in this manner?

It could do that part. Making the decision about what to buy, how much of it, and for whom etc would be the types of non-trivial decisions you need your sales AI to make better than a person

The F1 tickets are about access. Your pal the sales dude buys you scotch and yaks about SAP.

You're not gonna replace people who make phone calls all day without something more or less indistinguishable from strong AI. Nobody wants robo-calls.

You don't need chatbot tech to be indistinguishable from strong AI, just indistinguishable from an average salesperson. Like babyshake, I firmly believe that this will happen one day. 10 years? 50 years?

FYI for you and other readers, sales is actually hard. Low barrier to entry followed by brutal survival of the fittest. Don't meet your quota? Fuck you, you're fired.

People who survive for decades of that aren't trivially replaced.

Or as was put to me once, sales managers have no problem firing people. And that includes people who just had the bad luck to have their supporter at some company leave and be replaced by someone who favors a competitor. Or a project just gets put on hold because of budgets. I suspect many developers wouldn't appreciate being evaluated every quarter in no small part on factors that they have at best limited influence over.

You underestimate future AI tech (from 10-50 years from today) even more than you think I underestimate sales skills. The tech will completely blow your mind away, and you don't even see it coming ;)

At that point why would you even have AI sales?

If this tech is going to be so amazing, it'll be better served by being a procurement/buyer system.

"Hey AI, what CRM system should I use [based on all the knowledge you have of my business already]?"

Azure chatbot: Dynamics E7

correct, no need for salespeople

I suggest you actually try and sell a product once before making this kind of prediction.

OK, sure, but handwaving enough, that's true for almost every job that exists today.

You seriously think a salesperson can be replaced by a chatbot?

I feel like both sides think this,

Engineer - Our company is so good word of mouth should be enough to bring in sales, if we need any more lets look to AI for a solution.

Sales - We bring in the money, all engineers do is cost the company money, we could probably run the app/develop the product on a skeleton crew, can't wait for AI to make those jobs obsolete!

Truth - Both engineers and sales are hugely successful to any companies strategy and AI isn't taking anyone's(engineer or sales) jobs not for a long time :)

As a sales guy, I can assure you we value the development of good product over all else. It makes our jobs a hundred times easier.

Yea, but at most companies that doesn’t mean valuing all the devs ;)

Some boring business book:

Successful businesses must have 3 Successful divisions: good product, good sales and good finance/admin.

Willing to make a bet on that?

GPT-3 is great at sounding coherent, but much worse at actually reliably solving problems. I absolutely see GPT-3 like things aiding with sales decks, but... it won't replace the sales person.

There's also a large component of enterprise sales having a "trust" component where you know your counterparts. You're not spending approval-worthy amounts of money after only having talked to a chat bot.

> GPT-3 is great at sounding coherent, but much worse at actually reliably solving problems. I absolutely see GPT-3 like things aiding with sales decks, but... it won't replace the sales person.

Also, if your competitors are using GPT-3 to solve customer problems, having an actual, intelligent human can be a formidable differentiator in terms of customer experience.

> I absolutely see GPT-3 like things aiding with sales decks, but... it won't replace the sales person.

If there's a sales person whose job is to tailor sales decks to use cases, you might replace them.

If a big sale takes a team of people to execute, maybe AI cuts a few of them off. Maybe the client is still talking to a real sales person when they need to, but that sales person is able to multiplex to more clients because the time they need to put into each is reduced.

I think people often miss this aspect of "X technology will make this job obsolete"; it isn't "that occupation no longer exists" as much as "far fewer of that occupation are now necessary to do the work", and that change can have meaningful economic impacts. Farriers, blacksmiths and calligraphers still exist, but their work is no longer core to the function of anyone's economy.

Correct. The last professions in the world to be automated will be salesperson, politician, and plumber.

as a salesperson, I can confidently say that the change will be on the buyer side that eliminates the need for most salespeople. salespeople are already getting less and less time with prospects. Studies show salespeople only influence about 5% of the buying process. What comes next is automating recommendations for software products based on specific needs, buying criteria, etc.

Politicians have been automated through a combination of polling data and lobbyist checks since B. Clinton.

Are you saying B Clinton is not a human?

Define human.

It depends what the meaning of the word “is” is.

I think sex workers are up there too. Similarly to a good salesperson, the skill and human touch will probably always be a differentiator compared to a machine/AI, leaving at least a niche market for humans.

Customer support is largely useless today and can be replaced by machine. It only exists because average (dumb) people tolerate a human customer service rep saying "I'm sorry but you're screwed" better than a web page saying it.

Salespeople says BS to sign you up, only vaguely coherent, backed by some basic database facts about the product/price, so GPT can handle that. Think about Donald Trump: Huge fan base of paying customers, barely distinguishable from a chatbot.

Clearly you have never done or received customer support in a b2b environment. It can be as valuable or even more than the product itself.

Doubt it. Companies have been using chatbots with NLP for years. This only occasionally works to refer customers with technical support questions to relevant knowledge base articles. Mostly they're just a waste of time. Enterprise customers tend to have more complex support issues.

Chatbots exist to waste time to divert calls. There’s a high probability you’ll give up and not bother calling.

Only a moron would put them in front of a sale.

It will happen the day humans stop having emotions. So, never. Humans buy with emotions whether we like it or not. Emotions require another human to talk to. You cannot "Product Led or AI led" your way out of complex purchases.

Think about how complex it is to develop AI that can understand and participate in a human conversation at only a basic level.

Now increase that complexity by orders of magnitude, since we're talking about the conversational skillset that a successful salesperson would have.

Everything is possible over a long enough timeline, but when AI is advanced enough to function as a successful salesperson, there will be a large number of other jobs that AI will also be "replacing".

In other words, the timeline for what you're predicting is much longer than you think it is.

In other words, the timeline for what you're predicting is much longer than you think it is

Indeed. So many people I knew, thought full self driving, any weather, any road, fully AI, would be here by 2015. Then 20.

I doubt we will have it before 2050.

you're assuming sales conversations will be necessary in the future. I do not think that will be true for most purchases. It is much easier to match buyers with the right product based on their buying criteria. This is what buyers want anyway.

Well, Facebook is pretty good at automated advertising.

Those YouTube ads featuring pitchmen/people could be replaced by perpsnalizedt deepfake for race/gender/style that fits the target.

I think you are confusing a call centre employee, who blind calls via auto dialing, and reads a script, with a sales person selling a COTS, and doing inside sales.

There is zero comparison to this and a facebook ad, or targeted marketing.

GPT-3 ain’t buying me a coffee. They’re not inviting me to their Christmas party either.

B2B sales are generally relationship based, even inside sales. I don’t see a chatbot developing a relationship with a human customer.

This engineer thinks good engineers are much more fungible than good salespeople. The engineer's output is worthless unless someone else can be convinced to pay for it. It will be more difficult to automate the sales function than the engineering function.

Reply on news.ycombinator.com