Google "DeepMind" AI Chief Placed On Leave..

[chrisw]

As far as one can tell, the owners of Google, Alphabet, make 100 billion sales, 30 billion profits annually. Their entire future rests on staying cutting edge, hence the big spend buying up broad swathes of highly qualified talent, preferably all of it, I think is the line taken. If this costs 0.5 billion a year, well, they won’t be able to stop, or cut back, they have no choice but to continue paying out, anything else would be to give up. 0.5 billion out of 30 billion profits is nothing, they would be spending more, if they knew what to spend it on, which is basically the problem.

Well, taking into consideration that just this year alone DeepMind made more losses than the money Google paid to buy them, and also considering that Google probably already absorbed any useful expertise from this company, it really is questionable whether Google should continue servicing its debt.
These guys from DeepMind obviously have some deeper philosophy of "making the world a better place by using AI".
Google on the other hand is probably nothing but the worst morally corrupt money grabbing monopolist company in IT in the world. Considering this, I really don't see how DeepMind can continue existing inside Google much longer.

google doesn’t have any debt to service. you mean “carry on shelling out for the huge wages bill”. They don’t have any choice. Only choice is to rationalise, get rid of the dross (usually half of all programmers are lying to their managers/employers and do nothing constructive) and manage what they’ve got, better.

Well servicing losses in the correct phrase since there is simply no way one makes 500 million loss by paying wages. DeepMind has 700 employees. So even if average salary is 300k, which is highly unlikely, that is only 200 million for salaries. So where did those 300 million go?

Instead of arguing with me, and being wrong, why don't you go look it up in the published accounts? You'll find I'm right.

Simple answer, financing the cost of dubious projects led by the very shady incompetent Mustafa Suleyman.

And sorry but you are wrong about that guy.

Since I didn't comment on him, the wrong seems to belong to you.

I said that pilots don't run airlines, bus drivers don't run bus companies and programmers don't run companies, or shouldn't. Running businesses well, is a thing in itself. Remind yourself what happened to Apple when they let the programmers run everything in 1990s.

You can't be a successful CPO of an essentially highly specialized research cutting-edge hi-tech company worth half a billion at age of 26 and in the same time be clueless about actual technology your company is developing.

With good advisers, you probably can. It does help not to be clueless, but what's important is the ability to hold overview and not be taken in by programmer-promises bullshit, I guess neither you, nor me, know how good the guy and his advisers was at that.

DeepMind is not selling stupid apps, advertising stuff or collecting and selling user data so that a shady guy like Mustafa Suleyman could be successful as CPO.

[jp]

Forbes:

In 2018, DeepMind spent $483 million (£398 million) on around 700 employees, up from $243 million (£200 million) in 2017. Other significant costs included technical infrastructure and operating costs. In addition, DeepMind spent $17 million (£14 million) on academic donations and sponsorships.

DeepMind also spent $12 million (£9 million) on construction and $1.2 million (£1 million) on furniture and fixtures.

Yann LeCun, chief AI scientist at Facebook, said in an interview last year that he does not think DeepMind has yet proved its worth to Google, adding that DeepMind is too isolated to have a significant impact on the tech giant. “I wouldn’t want to be in the situation Demis is in,” he said.

Former DeepMind employees talk about tensions between Suleyman and cofounder and chief executive Demis Hassabis; with Alphabet’s own AI lab, Google Brain; and pressure after a U.K. healthcare project led by Suleyman became embroiled in a privacy scandal.

DeepMind Health’s first NHS project—developing a patient-monitoring app with the Royal Free Hospital in London—turned into a PR disaster for the company after the U.K.’s data regulator ruled that the transfer of 1.6 million patient records to DeepMind was illegal.

[Milos]

Well servicing losses in the correct phrase since there is simply no way one makes 500 million loss by paying wages. DeepMind has 700 employees. So even if average salary is 300k, which is highly unlikely, that is only 200 million for salaries. So where did those 300 million go?

Instead of arguing with me, and being wrong, why don't you go look it up in the published accounts? You'll find I'm right.

and

In 2018, DeepMind spent $483 million (£398 million) on around 700 employees, up from $243 million (£200 million) in 2017. Other significant costs included technical infrastructure and operating costs. In addition, DeepMind spent $17 million (£14 million) on academic donations and sponsorships.

DeepMind also spent $12 million (£9 million) on construction and $1.2 million (£1 million) on furniture and fixtures.

So it is even worse than I thought. It's obvious that someone is taking fat bonuses and filling their pockets. And I bet these are not the programmers and researchers. You can't see who is taking and how much money in bonuses in published accounts, but Google certainly knows this and I bet that effectively removing this shady guy Mustafa from the position of the CPO has something to do with it.

With good advisers, you probably can. It does help not to be clueless, but what's important is the ability to hold overview and not be taken in by programmer-promises bullshit, I guess neither you, nor me, know how good the guy and his advisers was at that.

That's a really stupid logic. Following this logic, no one outside the company can't know how successful some executive of the company is, and also to take any position you don't need any qualifications, it's only important to have good advisers. Give me a break, you can do better argumentation than this. Seriously.

[chrisw]

Well servicing losses in the correct phrase since there is simply no way one makes 500 million loss by paying wages. DeepMind has 700 employees. So even if average salary is 300k, which is highly unlikely, that is only 200 million for salaries. So where did those 300 million go?

Instead of arguing with me, and being wrong, why don't you go look it up in the published accounts? You'll find I'm right.

and

In 2018, DeepMind spent $483 million (£398 million) on around 700 employees, up from $243 million (£200 million) in 2017. Other significant costs included technical infrastructure and operating costs. In addition, DeepMind spent $17 million (£14 million) on academic donations and sponsorships.

DeepMind also spent $12 million (£9 million) on construction and $1.2 million (£1 million) on furniture and fixtures.

So it is even worse than I thought. It's obvious that someone is taking fat bonuses and filling their pockets.

it’s not obvious, except to someone who only just read the accounting details. okay, so now at least you found some figures.
staff costs (they also say ‘administrative costs’ in the actual accounts, maybe you should read those?) are not the same as pre tax wages seen by employees. Employees cost over and above their actual salaries.

And I bet these are not the programmers and researchers.

well, I have direct experience of what sort of figures the boss staff expect for themselves in these sorts of cases, and while large, they are not usually so large as to make up that substantial a slug of “cost of staff”. Staff cost most of the staff costs. And you actually know accurately neither the staff numbers, nor their pay rates.

You can't see who is taking and how much money in bonuses in published accounts, but Google certainly knows this and I bet that effectively removing this shady guy Mustafa from the position of the CPO has something to do with it.

you mean you wildly speculate. actually you don’t know.

With good advisers, you probably can. It does help not to be clueless, but what's important is the ability to hold overview and not be taken in by programmer-promises bullshit, I guess neither you, nor me, know how good the guy and his advisers was at that.

That's a really stupid logic. Following this logic, no one outside the company can't know how successful some executive of the company is,

well, random Miloses don’t know very well when they didn’t even read the accounting before opening mouth. and random Milos don’t know now either, just speculating based on some random Milos world view.

and also to take any position you don't need any qualifications,

who said that? straw man.

I said that to run a large bus company you don’t need to be a bus driver. Nor to run a large AI company do you need to be a programmer. Preferably not in either case. Viz Apple 1990s.

it's only important to have good advisers. Give me a break,

The problem here is that your idea of how companies run doesn’t scale. Beyond a certain size everything is different. That’s why little transport companies may well be run by a truck driver, but big transport companies are not.

you can do better argumentation than this. Seriously.

Thanks.

[noobpwnftw]

Isn't it ironic that our world leading AI company is having problems with overpriced human laboring?

[jp]

Telegraph:

One route to profitability could come through its new health ventures. Protein folding, the process by which chains of protein building blocks fold over each other to form 3D structures, has recently become a point of obsession for Hassabis - and for good reason.

Professor Michael Sternberg, the director of the Centre for Integrative Systems Biology and Bioinformatics at Imperial College, agrees.

DeepMind’s “performance at CASP13 was impressive,” he says. “I think people were taken aback by how from starting from nowhere, they really got right to the front.”

Turning DeepMind’s new protein folding ability into a source of revenue is somewhat of a mixed bag for the time, however.

“There quite clearly are major implications” from developments in protein folding, says Professor Sternberg. But there are alternative options to DeepMind’s system, many of which have been released for free online.

[jp]

Srivas Chennu, The Conversation, Dec. 2017:

DeepMind’s collaboration with London’s Royal Free hospital led to the NHS trust violating the UK’s data protection laws.

The Information Commissioner’s Office (ICO) found that Royal Free’s decision to share 1.6m personally identifiable patient records with DeepMind for the development of Streams – an automated kidney injury detection software – was “legally inappropriate”. DeepMind wasn’t directly criticised by the ICO.

Personal records included patients’ HIV-positive status, as well as details of drug overdoses and abortions. Royal Free’s breach generated considerable media attention at the time, and it means that DeepMind’s latest partnership with an NHS trust will be scrutinised carefully.

It will be working with Cancer Research UK, Imperial College London and the Royal Surrey NHS trust to apply machine learning to mammography screening for breast cancer.

Given DeepMind’s continued collaboration with the NHS on a range of research, citizens are rightly concerned about how private corporations might exploit the data they have willingly shared for publicly funded work.

Few details about the Royal Surrey research project – which is in the early stages of development – have been released, but it’s likely that DeepMind will focus on applying deep neural networks for scanning mammogram images to automatically identify signatures of cancerous tissue. This approach would be similar to its Moorfields Eye Hospital project, where DeepMind is building automated machine learning models that can predict macular degeneration and blindness from retinal scans.

[jp]

Forbes, Nov. 2018:

... DeepMind confirmed ... that its DeepMind Health unit is being handed over to Google following a report from CNBC's Christina Farr. The news has reignited concerns about patient privacy.

Officially announced in February 2016, DeepMind Health started out with around 15 people that were led by DeepMind cofounder Mustafa Suleyman, a left-wing activist whose mother was a nurse. The unit said it wanted to help the UK's National Health Service by providing apps and services.

The first Streams data-sharing agreement was in fact deemed to be illegal in July 2017 by UK data regulator Elizabeth Denham. At the time, she said the transfer of 1.6 million patient records from the Royal Free to DeepMind failed to comply with the Data Protection Act, adding that patients "would not have reasonably expected" their patient records to be used to develop an app.

DeepMind set up an independent review panel to scrutinise its work with the NHS, which also includes a partnership with Moorfields Eye Hospital to assess whether AI can be used to spot early signs of eye disease, and another with University College London Hospitals NHS Foundation Trust to explore the potential benefits that AI technology could have in planning treatment for patients suffering from head and neck cancers.

The review panel expressed concerns earlier this year that DeepMind might one day use its access to data to "drive monopolistic profits" for Google parent company Alphabet.

This panel is now going to be shut down, according to The Financial Times, which cites one person close to the situation, and CNBC.

Initially, DeepMind pledged to keep NHS patient records away from Google and it's repeatedly said it will never seek to commercialise patient data. "NHS patient data will only ever be used to improve hospital care," DeepMind previously said.

But the fact that highly sensitive patient records — containing addresses, prescriptions, and HIV statuses, among other things — are moving a step closer to Google itself has raised alarm bells.

Julia Powles, a research fellow focusing on law and technology at New York University, wrote on Twitter. "DeepMind repeatedly, unconditionally promised to *never* connect people's intimate, identifiable health data to Google. Now it's announced...exactly that. This isn't transparency, it's trust demolition."

[jp]

TechCrunch:

Google completes controversial takeover of DeepMind Health

The hand-off of DeepMind Health to Google, which was announced just over a year ago, means the tech giant is now directly providing software services to a number of NHS trusts that had signed contracts with DeepMind for Streams; as well as taking over several AI research partnerships that involve the use of NHS patients’ data to try to develop predictive diagnostic models using AI technology.

In its statement about the move yesterday, UCLH writes that “it was proposed” that its DeepMind research partnership — which is related to radiotherapy treatment for patients with head and neck cancer — be transferred to Google Health, saying this will enable it to “make use of Google’s scale and experience to deliver potential breakthroughs to patients more rapidly”.

It’s worth pointing out that the notion of “anonymised” high dimension health data should be treated with a healthy degree of scepticism — given the risk of re-identification.

Moorfields also identifies Google’s “resources” as the incentive for agreeing for its eye-scan related research partnership to be handed off, writing: “This updated partnership will allow us to draw on Google’s resources and expertise to extend the benefits of innovations that AI offers to more of our clinicians and patients.”

You could say Google is seeking to bundle access to the high resolution patient data that’s essential for developing health AIs with the provision of commercial digital healthcare services it hopes to sell hospitals down the line, all funnelled through the same Google cloud infrastructure.

Patient consent was not sought for passing medical records then; nor have patients’ views been consulted in the transfer of Streams contracts (and people’s data) to Google now.

And while — after it was faced with public outcry over the NHS data it was processing — DeepMind did go on to publish its contracts with NHS trusts (with some redactions), Google Health is not offering any such transparency on the replacement contracts that have been inked now.

[bob]

https://www.dailymail.co.uk/news/articl ... jects.html

"Over 500 Million Dollars in Losses"

"Co-founder of Google's artificial intelligence lab DeepMind takes leave of absence amid 'controversy over projects he's overseen' and company losses of $571 million
Mustafa Suleyman, 34, has taken a leave of absense from DeepMind, an artificial intelligence lab owned by Google's parent company Alphabet
Suleyman runs DeepMind's 'applied' division, which deals with health, energy and technology sectors
A spokesman said he was taking a 'time out', adding it was a mutual decision
It is expected that Suleyman, who is in his 30s, will return by the end of the year
Suleyman’s team had been responsible for finding ways for DeepMind to make money, working on a text-to-speech service for Google Cloud
However, earlier this month DeepMind’s losses rose 55 percent to $571 million
The company he founded may be owned by one of the world’s largest companies, however Suleyman has said ‘capitalism is a failing society’
His ambition with DeepMind is to ‘solve the world’s toughest problems’, making no mention of how such ventures could be profitable" (DailyMail)

"Suleyman founded DeepMind in 2010 alongside current Chief Executive Officer Demis Hassabis.

Four years later, Google bought out the company for $486 million, which led to an expensive race in Silicon Valley for specialists in the field of Artificial Intelligence.

Reports of Suleyman’s hiatus Wednesday sparked speculation of a rift between Google and DeepMind over how to commercialize their AI products.

Suleyman’s team had essentially been responsible for finding ways for DeepMind to make money, working on a text-to-speech service for Google Cloud and cutting Google’s data center cooling costs.

However, earlier this month, the Financial Times reported that DeepMind’s losses had risen 55 percent to $571 million, credited largely to a series of aggressive bids made by Suleyman to secure the world’s brightest artificial intelligence minds under DeepMind's payroll."

Wow...Well my Friends this doesn't sound too good at all... How much did the Alpha Zero Project Cost I wonder?

I never even understood in the first place, how a guy that knows practically nothing about programming or actual AI research could be a CPO of a leading AI company.
Demis Hassabis and especially Shane Legg (as a curiosity I actually attended his PhD defence ) are really bright guys and world-class experts in the field, but this Mustafa Suleyman looks like nothing but a totally shady guy.

that’s like telling us a bus company has to be run by someone who can drive buses, or an airline company has to be run by a pilot. In fact there is probably nothing worse that programmers running software companies.

“looks like shady” is chronically subjective comment. I’ld imagine there are far deeper reasons for this situation. Like, they’ve expanded very fast, taking on much talent, or all the talent even, Much of that will be dross, people who bullshit, and it takes time and skill to weed those out. Secondly, deepmind is trying to be a university, basically, but also it is supposed to do stuff, so it has software engineers who were never intended to be creative, and creative brain people, who are intended to be creative. The former get told what to do, the latter are supposed to b self-driving. When you’re burning money at 0.5 billion, there’s going to be pressure to rationalise and weed out the dross. There’s also the problem of “what to do” with all this talent. I doubt there’s enough be generated at various stages of advancing AI, so they hit bottlenecks and ceilings after a while. All this requires management (definitely not by programmers btw). I would guess it’s an over-growth rationalisation and reorganization crisis which will be overcome, because it has to be overcome and because Google have got so much money, they need to be deploying it. Deepmind would have to become utterly hopeless before they’ld even consider shelving it. What else is google to do, but employ all the smart researcher class and hope they find things before FB and all the other cash rich software AI media companies beat them to it.

Actually I believe that _IS_ correct, you DO need experience in the field. If you hire someone to run (say) Trailways (passenger bus company in the US) that knows nothing about busses, how is he going to succeed? 5 years worth of learning how to schedule, market, purchase, etc? Probably not the best way to start a company. Not saying he needs to be a neural net expert. But he certainly needs a lot of experience with a software development company, and such people usually come up from the ranks of programmers, which helps them understand that which they will be managing.