Connect with us

Uncategorized

OpenAI’s DALL-E creates plausible images of literally anything you ask it to

Published

on

OpenAI’s latest strange yet fascinating creation is DALL-E, which by way of hasty summary might be called “GPT-3 for images.” It creates illustrations, photos, renders, or whatever method you prefer, of anything you can intelligibly describe, from “a cat wearing a bow tie” to “a daikon radish in a tutu walking a dog.” But don’t write stock photography and illustration’s obituaries just yet.

As usual, OpenAI’s description of its invention is quite readable and not overly technical. But it bears a bit of contextualizing.

What researchers created with GPT-3 was an AI that, given a prompt, would attempt to generate a plausible version of what it describes. So if you say “a story about a child who finds a witch in the woods,” it will try to write one — and if you hit the button again, it will write it again, differently. And again, and again, and again.

Some of these attempts will be better than others; indeed, some will be barely coherent while others may be nearly indistinguishable from something written by a human. But it doesn’t output garbage or serious grammatical errors, which makes it suitable for a variety of tasks, as startups and researchers are exploring right now.

DALL-E (a combination of Dali and WALL-E) takes this concept one further. Turning text into images has been done for years by AI agents, with varying but steadily increasing success. In this case the agent uses the language understanding and context provided by GPT-3 and its underlying structure to create a plausible image that matches a prompt.

As OpenAI puts it:

GPT-3 showed that language can be used to instruct a large neural network to perform a variety of text generation tasks. Image GPT showed that the same type of neural network can also be used to generate images with high fidelity. We extend these findings to show that manipulating visual concepts through language is now within reach.

What they mean is that an image generator of this type can be manipulated naturally, simply by telling it what to do. Sure, you could dig into its guts and find the token that represents color, and decode its pathways so you can activate and change them, the way you might stimulate the neurons of a real brain. But you wouldn’t do that when asking your staff illustrator to make something blue rather than green. You just say, “a blue car” instead of “a green car” and they get it.

So it is with DALL-E, which understands these prompts and rarely fails in any serious way, although it must be said that even when looking at the best of a hundred or a thousand attempts, many images it generates are more than a little… off. Of which later.

In the OpenAI post, the researchers give copious interactive examples of how the system can be told to do minor variations of the same idea, and the result is plausible and often quite good. The truth is these systems can be very fragile, as they admit DALL-E is in some ways, and saying “a green leather purse shaped like a pentagon” may produce what’s expected but “a blue suede purse shaped like a pentagon” might produce nightmare fuel. Why? It’s hard to say, given the black-box nature of these systems.

But DALL-E is remarkably robust to such changes, and reliably produces pretty much whatever you ask for. A torus of guacamole, a sphere of zebra; a large blue block sitting on a small red block; a front view of a happy capybara, an isometric view of a sad capybara; and so on and so forth. You can play with all the examples at the post.

It also exhibited some unintended but useful behaviors, using intuitive logic to understand requests like asking it to make multiple sketches of the same (non-existent) cat, with the original on top and the sketch on the bottom. No special coding here: “We did not anticipate that this capability would emerge, and made no modifications to the neural network or training procedure to encourage it.” This is fine.

Interestingly, another new system from OpenAI, CLIP, was used in conjunction with DALL-E to understand and rank the images in question, though it’s a little more technical and harder to understand. You can read about CLIP here.

The implications of this capability are many and various, so much so that I won’t attempt to go into them here. Even OpenAI punts:

In the future, we plan to analyze how models like DALL·E relate to societal issues like economic impact on certain work processes and professions, the potential for bias in the model outputs, and the longer term ethical challenges implied by this technology.

Right now, like GPT-3, this technology is amazing and yet difficult to make clear predictions regarding.

Notably, very little of what it produces seems truly “final” — that is to say, I couldn’t tell it to make a lead image for anything I’ve written lately and expect it to put out something I could use without modification. Even a brief inspection reveals all kinds of AI weirdness (Janelle Shane’s specialty), and while these rough edges will certainly be buffed off in time, it’s far from safe, the way GPT-3 text can’t just be sent out unedited in place of human writing.

It helps to generate many and pick the top few, as the following collection shows:

AI-generated illustrations of radishes walking dogs.

The top 8 out of a total of X generated, with X increasing to the right.

That’s not to detract from OpenAI’s accomplishment here. This is fabulously interesting and powerful work, and like the company’s other projects it will no doubt develop into something even more fabulous and interesting before long.

Lyron Foster is a Hawaii based African American Musician, Author, Actor, Blogger, Filmmaker, Philanthropist and Multinational Serial Tech Entrepreneur.

Continue Reading
Comments

Uncategorized

Unmind raises $47M for a platform to provide mental health support in your workplace

Published

on

Mental health has been put into the spotlight in a big way in recent times. For many of us, our lives and lifestyles have changed massively in the last year, and alongside that, we’re collectively facing pandemic-fueled mortality on a global scale in a way that hasn’t existed for generations, a perfect storm of sorts that has inevitably had an impact on our state of mind and our moods.

Today a startup that has built a platform to help people think about and respond to this situation is announcing a big round of growth funding, specifically to help address all of this and how it plays out in one of the more stress-inducing aspects of our life — our workplaces.

Unmind — a London startup that has built a mental health app for the workplace — has raised $47 million, a Series B that it will be using to continue investing in its research and development and also to expand its business reach. The funding is being led by EQT Ventures –- a very active investor at the moment in UK growth rounds — with participation also from Sapphire Ventures and previous backers Project A, Felix Capital, and True.

The core of Unmind’s service is an app built around a set of questions to help employees explore their own states of mental health, which could include depression, anxiety, insomnia, and a host of other manifestations. It provides advice and content to begin addressing the results of that — exercises, advice, podcasts, links for further reading, and links to seeing further help from professionals (not more machine interfaces, but humans). It also provides a service to the employers, sharing anonymized data from the app with them so that they, too, can consider how better to respond to their employees’ needs.

The app has seen some notable traction especially in the last year, a time when the conversation about mental health has become much more commonplace and critical, given the environment we’ve been living in.

Unmind does not disclose user numbers, nor how they have grown, but it tells me that uptake and adoption of its app ranges from 15% to over 60% of an organization’s workforce (this varies by size, and the emphasis that the organization itself puts on using the app, among other things). It said that of those employees who are using Unmind, 88% have said they experience an improvement in mental wellbeing, work, or relationships, while 92% report higher confidence, awareness, and understanding of mental health.

The company also said that revenues grew by more than 3x in the last 12 months. Meanwhile, its customers include major retailers like John Lewis and M&S, high street bank TSB, Uber, Samsung, Virgin Media, British Airways and Asos — a list of companies that have strong degrees of customer service around them, have been greatly impacted by the lockdowns, and you can imagine must have a lot of people working in them pretty stressed out as a result of being on the front lines of interfacing with a stressed-out wider population of consumers.

The company was co-founded by Dr Nick Taylor, who previously had been a clinical psychologist and worked for years in mental health care (and before that was a classically-trained singer), who said he came up with the idea after feeling like he was seeing too many people only for the first time at a stage when their issues were already very advanced.

“I kept encountering the same frustration time and again: I wish I’d met this person six months ago,” Taylor said in an interview.

As with all kinds of preventative healthcare, it’s always better to identify and work on issues before they grow big and more urgent, and so he set out to think about how one might approach the concept of a preventative check-up and check-in for mental health.

The workplace is not a bad place to base that effort. Not only is it often a source of stress for people, but it’s a regular place for them to be every day so creating a way of assessing mental health through that implicitly creates a kind of routine to the effort. It also potentially means a closer connection to the employer to work on issues more collectively when and if they emerge, in a way that the employer might not do (or ever discover) through other means.

The connection between work and mental health is a longstanding one but has perhaps been proven out more than ever before in the last year.

“I didn’t know what would happen with mental health during Covid,” Taylor recalled. “I actually wondered if it would be demoted,” given all of the other conflicting priorities. “But the prevalence of mental illness has escalated. It’s out of control. And in the workplace, it’s a leading cause of absenteeism and turnover.” And given how full-on everything has become, including likely more hours spent working since now it all has merged with our home lives, we all know (and may well be among) many people who are feeling incredibly burned out right now.

Taylor said that in fact quite the opposite has happened to his early skepticism: mental health has become front of mind, “and the shackles of stigma are falling away.”

This is part of what has really caught the eye of investors: technology that is not just effective, but very relevant to right now. “It is now universally recognized that our Mental Health is as important if not more important than our physical health – but has long been neglected. That is now changing rapidly,” said Alastair Mitchell, a partner at EQT Ventures. “As a result there has been a massive rise in the popularity of consumer mental health apps which is now being matched by surging demand from employers and employees for the same in the workplace. Unmind is the leading mental health app for the enterprise and we are so excited to work with Dr Nick and the team to support their scaling globally.” EQT is also a strategic investor, not just a financial one: it’s rolling out Unmind across its own workplace and its many portfolio companies.

Unmind, it should be noted, is not the only company that has identified this “opportunity,” if you could call it that. They include other startups like SF-based Ginger — which has also built a platform that partners with employers, but also healthcare providers and other stakeholders, to help people identify and manage their state of mind. Ginger has been well-capitalised over the years. Others in the same space include Welbot in New York, Spill also out of London and a host of others providing different aspects of mental wellness like Calm and Headspace, the meditation apps.

I’m inclined to think that, given the size of the problem and that mental health should not be a bunfight but something that takes a village to address, the key will be in how each company approaches its remit, and how people respond to it, and whether what people do ultimately use results in better bridges for employees to getting the help and peace they need, whether it’s from the app or a professional.

“We have a responsibility to connect with our mental health in the same way that we do when it comes to healthcare,” Taylor said, likening the effort to how it takes a number of skill sets sometimes to work on the complexities of a health issue. “Great healthcare integrates across a number of systems.”

Continue Reading

Uncategorized

Dutch startup QphoX raises €2M to connect Quantum computers with a Quantum modem

Published

on

When eventually they become a working reality, Quantum computers won’t be of much value if they simply sit there on their own. Just like the internet, the value is in the network. But right now there’s scant technology to link these powerful devices together.

That’s where QphoX comes in. Thus Dutch startup has raised €2 million to connect Quantum computers with a ‘Quantum modem’.

The funding round was led by Quantonation, Speedinvest, and High-Tech Gründerfonds, with participation from TU Delft.

QphoX aims to develop the Quantum Modem it created at Delft University of Technology (TU Delft) into a commercial product. This networks separate processors together, allowing quantum computers to scale beyond 10’s or 100’s of qubits. Look out for the Singularity folks…

Simon Gröblacher, CEO and co-founder of QphoX told me: “It is the exact same thing as a classical modem except for quantum computers, so it kind of converts electrical and microwave signals to optical signals coherently, so you don’t do any of the quantum information in the process. It then converts it back so you can really have two quantum computers talk to one another.

I noted that there’s more than one type of quantum computer. He countered “We are in principle agnostic to what kind of quantum computer it is. All we do at the moment is we focus on the microwave part, so we can work with superconducting qubits, topological qubits etc. We can convert microwaves to optical signals and they can talk to each other. Currently, the only competitors I know are all the in the academic world. So this is we’re the first company to actually starts building a real product.”

Rick Hao, Principal with Speedinvest’s Deep Tech team, added: “ We want to invest in seed-stage deep technology startups that shape the future and QphoX is well-positioned to make a major impact. Over the next couple of years, there will be rapid progress in quantum computers. Quantum Modem, the product developed by QphoX, enables the development of quantum computers that demonstrate quantum advantage by combining separate quantum processors.”

Continue Reading

Uncategorized

UK fashion portal Lyst raises $85M in a ‘pre-IPO’ round, reportedly at a $500M valuation

Published

on

E-commerce continues to be a huge focus for investors watching consumer behavior and spending patterns in the wake of the Covid-19 pandemic. In the latest development, UK startup Lyst, a portal for high fashion brands and stores to sell directly to users, has picked up $85 million, in what the startup is describing as a ‘pre-IPO’ round.

The news comes as the company says that it has now grown to 150 million users browsing and buying from a catalog of 8 million products from 17,000 brands and retailers.

List said that gross merchandise value in 2020 was over $500 million, with new user numbers growing 1100% growth in new users. GMV has definitely been accelerating. Lyst has been around since 2010 and said today that lifetime GMV is more than $2 billion.

“Lyst is rapidly becoming a fashion category leader, which hundreds of millions of fashion lovers rely on to decide what to buy. While our app and website already enjoy very large audiences in the USA & Europe, fashion e-commerce remains under-penetrated in general, with huge growth potential globally. We’re excited to use this raise from top-tier investors to continue personalising the fashion shopping experience to each of our millions of customers, while helping our partner brands thrive,” said Chris Morton, Lyst’s CEO and founder, in a statement.

We have contacted the company to ask about the timing and location for a public listing and while it has not commented, we understand that London or New York would be the most obvious locations for a listing, which is not likely to be for another year or even three.

For now, Lyst has disclosed that investors in this latest injection include funds managed by Fidelity International, Novator Capital, Giano Capital and C4 Ventures, as well as a mix of financial and strategic previous backers Draper Esprit, 14W, Accel, Balderton Capital, Venrex and LVMH. Carmen Busquets — a strategic advisor to the company who co-founded Net-a-Porter, one of Lyst’s competitors in the space — also increased her investment in the company with this round, the company said.

Lyst is not disclosing its valuation but PitchBook notes that with this round, it is $500 million post-money. (We’ve also asked the company to confirm whether this is an accurate figure.) Sky News, where the funding news was leaked last night, did not have a valuation figure.

For some further comparison and context, though, Farfetch, another competitor in the same space as Lyst, listed publicly some years ago and currently has a market cap of $14.4 billion. And more generally, there is a lot to play for here online, not just against other pure-play fashion portals, but also standalone retailers, marketplaces like Amazon, and increasingly social media apps like Instagram, TikTok and Snapchat, which are all looking at how they can better capitalize on how their platforms are already being used quite aggressively and widely for social commerce.

Social media sites would be an ironic but perhaps very unsurprising competitor for Lyst, which started life as a pioneer in the concept, creating a way for people to follow influential high fashion brands and influencers on its platform — who were not actually called “influencers” at the time, but curators and bloggers (the more things change, eh?) — and get alerts when items would be posted by them for sale.

People might have originally been very skeptical about how well high fahion (read: expensive, sometimes esoteric) might play over screens, but over time Lyst and the others in the same proved it all out in spades, raising successive rounds over time to back up its premise. Balenciaga, Balmain, Bottega Veneta, Burberry, Fendi, Gucci, Moncler, Off-White, Prada, Saint Laurent and Valentino are among the brands that appear on Lyst today.

Over the years, more variations and competitors have presented themselves, but the salient fact remains that high fashion has a huge target audience delivered in the right way, and that is something that investors, brands, influencers, and these marketplaces themselves have all doubled down on in the pandemic.

It’s been a time when people who have not found themselves outright struggling financially (and there are lot of those, unfortunately), have instead found themselves with more disposable income since they went out and travelled significantly less than before. Fashion and buying goods for ourselves has become a form of escapism, and for those who get a lift out of the tree falling in the forest and being there to hear the sound, we can still put on the outfits, snap ourselves for our Stories, and exposure will still be ours.

“Lyst has made huge progress over the past year with its industry leading app for the fast- growing online luxury fashion market – a trend which looks set to continue as consumers retain their newfound digital habits, and demand for fashion rises further post-pandemic. In recent years we have seen other high-growth fashion tech businesses taking the next step, and we believe Lyst is well positioned to capitalise on this market momentum. Draper Esprit has backed Lyst since Series A and we believe this latest round sets the business up for an exciting next phase,” said Nicola McClafferty, a partner, Draper Esprit, in a statement.

Lyst also announced a few appointments to firm up its executive bench in the lead-up to its next steps as a company. Mateo Rando previously at Spotify, is joining as chief product officer to focus largely on Lyst’s popular mobile app. And Emma McFerran, formerly general counsel and chief people officer, is stepping up as COO and a new board member.

Continue Reading

Trending