AI Startup Claims To Automate App Making But Actually Just Uses Humans (theverge.com) 38
Engineer.ai, an Indian startup claiming to have built an artificial intelligence-assisted app development platform, is not in fact using AI to literally build apps, according to a report from The Wall Street Journal. Instead, the company, which has attracted nearly $30 million in funding from a SoftBank-owned firm and others, is reportedly relying mostly on human engineers, while using hype around AI to attract customers and investment that will last it until it can actually get its automation platform off the ground. The Verge reports: The company claims its AI tools are "human-assisted," and that it provides a service that will help a customer make more than 80 percent of a mobile app from scratch in about an hour, according to claims Engineer.ai founder Sachin Dev Duggal, who also says his other title is "Chief Wizard," made onstage at a conference last year. However, the WSJ reports that Engineer.ai does not use AI to assemble the code, and instead uses human engineers in India and elsewhere to put together the app. When pressed on how the company actually employs machine learning and other AI training techniques, the company told the WSJ it uses natural language processing to estimate pricing and timelines of requested features, and that it relies on a "decision tree" to assign tasks to engineers. Neither of those really qualify as the type of modern AI that powers cutting-edge machine translation or image recognition, and it does not appear that any kind of AI agent or software of any kind is actually compiling code.
Recursion (Score:2)
Nothing new to programmers who use Emacs (Score:4, Insightful)
The code is generated by a semi-sentient program with additional input from a human.
It can do anything! (Score:2)
the type of modern AI that powers cutting-edge machine translation or image recognition
I think you over-estimate the power of AI machine translation. It's all weak AI.
Strong AI: machines that can think like a human. We're not even close.
Re: (Score:2)
Same with self-driving cars by the way. People love to dream up scenarios that would require mind reading. Yeah, but, it won't actually matter.
Re: (Score:2)
No, I think Machine AI is useful and will become pretty good. You can always dream up corner cases it won't handle without having general knowledge
Machine AI is already useful. A computer doesn't need AI at all to be useful. Sorting is useful. A* search was discovered as weak AI, and it is immensely useful.
The point I'm trying to make is that strong AI is different than weak AI, not that weak AI is useless (I think I wasn't clear because you misunderstood me). Reporters and investors confusing the difference between strong AI and weak AI is what leads people to apologize to Alexa so she won't enslave them in the robot revolution. Which makes no sens
Re: (Score:2)
Re: (Score:2)
... leads people to apologize to Alexa so she won't enslave them in the robot revolution. Which makes no sense if you understand the technology behind Alexa.
Funny, didn't know people did that. Though doubt that many does that in order not to make the machine angry. Those devices are made to be interacted with as similarly as possible to how you interact with another human. Most of us apologize to other humans now and then, and conversation often require little focus, so it makes sense that you sometimes forget for a moment that you are talking to a can.
Now that I think of it, perhaps saying "sorry" or "thanks" to your virtual assistant could be useful feedback
Re: (Score:2)
Strong AI isn't "thinking like a human". Strong AI is coming remotely close to thinking like a human.
Weak AI is what we have today : nothing close to human.
weak vs strong (Score:2)
from what I'm remembering from my studies (but it's not my main field, as my nickname suggest. Cue in Dr Bones "I'm a doctor, Jim, not a...").
Weak AI: machine learning technique specially taylored to a specific problem.
Think heuristic engine doing min-maxing on a Chess play. You won't be able to throw the same software at voice recognition problems.
Strong AI: a theoretical smart intelligence which can be applied to anything.
Think all the robot assistants in SciFi movies that otherwise behave like a human as
Re: (Score:2)
Why do people believe this? (Score:2)
I always wonder why people believe in these AI claims. Have you EVER seen a computer do something even remotely intelligent? And if you DID figure out how to build an AI would you waste it on building...apps? Some people believe computers are magic I guess.
Re: (Score:2)
I always wonder why people believe in these AI claims. Have you EVER seen a computer do something even remotely intelligent? And if you DID figure out how to build an AI would you waste it on building...apps? Some people believe computers are magic I guess.
I am in agreement with you 100%. Except I know that a computer beat human players in the game of "go". A game invented in china over 2000 years ago. It watched a bunch of real games played by humans, then it "played" a bunch of games against itself ( ala war games style) and it was then able to beat human players using techniques that no human has ever considered.
There might be something there.
https://fortune.com/2016/03/12... [fortune.com]
Maybe some specific things , for now, they can make magic.
Re: (Score:2)
I am in agreement with you 100%. Except I know that a computer beat human players in the game of "go".
The Go computer won because of its ability to search through millions of positions very, very quickly. As computers got faster, they were able to search through more positions faster and faster, and they got better and better compared to humans. They were on a trend to beat humans as CPU speeds increased. Google threw an order of magnitude more CPU power at the problem than anyone else did, and got there 10 years sooner than the trend. But it's worth remembering that at its core, AlphaGo is primarily a tre
Re: (Score:2)
Strange, I always though AlphaGo etc. were deep learning neural networks, and not 'standard' tree exploring algorithms.
Re: (Score:2)
Strange, I always though AlphaGo etc. were deep learning neural networks, and not 'standard' tree exploring algorithms.
Read my post again. Where do you think they are using the deep learning neural network? LTR
Re: (Score:2)
They actually do both. Or tripple.
2 deep learning neural networks.
One for judging positions on the board.
One for proposing possible moves.
And a standard, as you say, tree pruning, deep search algorithm to judge the two networks above.
Re: (Score:2)
Re: (Score:3)
Some people believe computers are magic I guess.
Most people are not computer engineers nor computer scientists, ergo... any sufficiently advanced technology is indistinguishable from magic. (Kudos to Clarke.)
Re: (Score:2)
Oh, come on!
For once, they are actually using intelligence in their AI solution, and you are still complaining?
for money ! (Score:2)
They are simply playing the "Venture Capital " Bingo:
Trying to cram in as many over-hyped buzzwords as possible in order to attract angel investors.
- AI is a popular buzzword in the news.
- Apps are a fast growing market.
Combine them and hope for a big Ka-ching!
(the only reason they didn't used "3d printers" and "blockchain" in their blurb is because they probably didn't imagine a good justification to cram them too.
Otherwise they would have become instant bazillionairs).
A.I. vs a.i. (Score:2)
So they are using a.i. (actual intelligence) instead of A.I. (Artificial Ignorance).
Welcome to programming 101.
Nope.. ASS. (Score:2)
That are using ASS (Actual Sentient Stupidity) to get big piles of VC based on lies.
In other words, they are following the standard tech startup process.
Almost (Score:2)
almost.
They didn't manage to put "drones" and "cloud" in their blurb to win the bullshit-buzzword bingo, they'll be missing out on some VC money.
Re: (Score:2)
Pretty much. Even if AI becomes "competent" someday. Humans still told it how.
Real AI (Score:1)
OK - so the machine determines what it is that needs to be done and then outsources it to humans to do the work. Welcome to the brave new world.
Cultures (Score:2)
Re: (Score:2)
Ever heard of General Electric?
Re: (Score:2)
*cough* Theranos *cough*
There are hucksters everywhere there are suckers.
Re: (Score:2)
Well they promis to create apps for your plain vanilla business cheap and fast. And they deliver on that. How they do it.... yes, they say they use AI to do it. But without any useful definition of AI they can have the company spam filter run by a neuronal network and they are using it.
Actual Intelligence (Score:1)
ducklife (Score:1)
We all know (Score:2)
AI = A lot of Indians
This has been the case for many years. Still not at the level of western humans.
open source already is a collective AI (Score:1)
a good programmer is "lazy": we will rather create an abstraction layer than repeat ourselves more than 3 times. Often these abstractions are reusable in other software. Open software is the next level of "laziness" where programmers understand that sharing a solution to common problems is the most intelligent thing to do. Github adds an extra social layer where the best ideas bubble to the top.
Neural Networks (Score:2)
Neither of those really qualify as the type of modern AI that powers cutting-edge machine translation or image recognition, and it does not appear that any kind of AI agent or software of any kind is actually compiling code.
Neither do the current re-designs of the good-old 80s neuronal networks.
Even though I have to admit with the available processing power it's amazing what they can do today.
An old Soviet era joke comes to new merits (Score:2)
Glorious Soviet Union developed peeling machine for potato. Was a big machine, but you throw potato in on top, machine works for seconds, then potato comes out from bottom.
At demonstration, honorable comrade from party was presented glorious development of Soviet ingenuous engineering effort. Honorable comrade was very impressed, took bag of potatoes and dumped it all into machine. Machine rustled, then side door opened and old babushka looked out, whining "Niet, not so much, I am alone in here!"
Test Tube Babies (Score:2)
Sudden Twist: It is AI. The programmers were artificially created in a laboratory. They have no parents except for their lab caretakers.