There is no cure for Alzheimer's, even though it's one of the most common diseases in older people.
It takes a lot of time and money to produce even the most everyday medicines, let alone a cure for something like Alzheimer's.
According to the Association of the British Pharmaceutical Industry, it costs £1.12 billion and up to 12 years to develop a single drug.
Imagine if artificial intelligence could speed that process up. We might start seeing cures for Alzheimer's and even rarer diseases being developed much faster.
That's what British startup BenevolentAI wants to do. The company uses artificial intelligence (AI) to help researchers identify possible cures more quickly.
Eventually, the company wants to sell AI-developed drugs directly, founder Ken Mulvany told Business Insider in an interview. He predicted BenevolentAI would be selling its own drugs "in the next four years."
BenevolentAI uses artificial intelligence to make connections that humans can't
Drug discovery is a complicated process, but it begins when researchers try to understand what factors (or "mechanisms") cause a disease.
They then use existing research to hypothesise how a particular compound might affect that mechanism and, in turn, cure the disease. Once they've tested their hypothesis, those compounds may be taken into development.
BenevolentAI's tech comes in at that educated guess stage.
Its AI processes academic literature, studies, and other data about particular diseases, and uses this to help researchers come up with a hypothesis.
"Today, we have so much information," Mulvany told Business Insider. "90% of the world's information has been produced in the last two years, it's a tremendous amount of data. As humans we cannot assimilate that. There's so much out there, you need a system that can relate all of these things together."
BenevolentAI's technology can identify correlations that a human researcher would never think to look at, Mulvany said. Researchers who aren't necessarily specialists in a particular area can also use Benevolent AI's systems to interrogate a disease, he added.
The company claims its processes mean researchers can come up with new hypotheses that were "previously impossible" because of the sheer amount of data its AI can process.
BenevolentAI has drugs in development
Unlike other AI companies like DeepMind, BenevolentAI doesn't publish academic papers about its technology. But there are signs it's serious.
Mulvany was previously CEO of another drug development company Proximagen, which he sold for around £223 million to US drugmaker Upsher-Smith.
At Proximagen, he said, it took 10 years to get 15 drug candidates to development. At BenevolentAI, the company has 24 drug candidates in just four years.
And the startup signed an $800 million (£624 million) deal in 2014 to hand over two Alzheimer drug targets to an unnamed US company for development, Mulvany said. This means Benevolent AI isn't working on Alzheimer's drugs itself any longer, but will take a cut from the profits if its drugs are developed and sold.
Mulvany said Benevolent AI would develop and sell drugs for rare diseases itself, but partner larger firms for more common illnesses like Alzheimer's. It's already working on drugs for ALS, the neurological disease that became well-known after the Ice Bucket Challenge.
Along with SAP, BenevolentAI became one of the first companies to use Nvidia's DGX-1 supercomputer, which helps companies train AI software.
The company told Business Insider it has raised $87.7 million (£68 million) in total from investors including Woodford Investment Management, Lundbeck, Lansdowne Partners, and Upsher Smith, at a valuation of $1.78 billion (£1.4 billion).
Recent filings with Companies House show that Horizons Ventures advisor Bart Swanson sits on the company's board, and a long list of shareholders including Bruce Castle, founder of Canadian investors StoneCastle, and Richard Farleigh, a high-profile angel investor. The filings also show a £12 million loss for the 13 months to December 2015, on turnover of £1.3 million.
BenevolentAI's CEO said DeepMind didn't have a business model
YouTube/Royal Television Society
Mulvany sits on government advisory boards about AI ("It's mostly about jobs," he said) along with Demis Hassabis, the cofounder of DeepMind.
Mulvany has strong opinions about DeepMind, its acquisition by Google, and the London tech scene more widely.
"I like them," he said."I like the group. The acquisition has done a lot for London and the UK in general to bring the academic piece into focus. It's easier for us to attract [talent] to London than it would have been if Google hadn't made the acquisition."
In short, DeepMind made AI sexy.
But he added: "With no disrespect to DeepMind, they didn't have a business model. There are few places you can be acquired with no business model, and just [by saying] 'We do cool stuff.'"
He said DeepMind faced a big challenge by trying to "solve general intelligence," whereas BenevolentAI is "a very narrow AI company."
"This is what we're good at, and we're able to apply [our tech] today," he said. "We look forward to what they publish, when they publish it. But I don't know, and they don't know either, how long that's going to take. It could be five, 10, or 50 years."
Mulvany added that DeepMind isn't a rival. "DeepMind does healthcare, we do drug development," he said.
When it comes to talent, BenevolentAI more often competes against Calico, Google's life sciences division, and Human Longevity Inc., the human genotypes company cofounded by biotechnologist Craig Ventner.
BenevolentAI isn't going to sell like DeepMind
Mulvany said he doesn't want to sell BenevolentAI, and is looking to expand the company into material sciences.
He said British startups should stop selling out early, and said it was "necessary" to build sustainable companies in the UK.
"It's necessary to do that," he said. "I was very outspoken [at Proximagen] about UK technology ending up in the hands of US companies. And I remember saying, 'I'm not going to sell this company, because I'll end up selling it for far less than it's worth.' I was outspoken publicly, to the government, and then I ended up selling it. I had malaise of the infrastructure I needed to grow the business in this country."
He added: "I'm not going to do it again! There's a real opportunity to build a business in this country, and there's an opportunity for the country to lead this sector, and all the pieces are in place. We've got incredible scientists, incredible mathematicians, we're able to draw all these people into London."
- Google's CEO isn't worried about making money on its most futuristic products yet
- Amazon's new Echo camera is weird and a little creepy — but it hints at Amazon's master plan to rule tech
- Investors backed an AI startup that puts a doctor on your smartphone with $60 million