The AI revolution: Is Israel keeping up?

The fact that Israel does not have a GPU farm is equivalent to Israel not having a nuclear reactor in the ’50s, said Navan, co-founder and CTO Ilan Twig, speaking at artificial intelligence initiative The Institute.

The remark was made at the launch of The Institute – a collaborative effort to bolster Israel’s competitive edge in the field of AI. It delivers training programs for corporate executives and aims to give them the tools that they need to navigate changes made to the business world by AI. It also aims to serve as a hub for educating the next generation of professionals.

The initiative is led by venture capital group Group 11 in partnership with Ben-Gurion University of the Negev and prominent global corporations such as Dream and Microsoft. Among participating companies is software company NVIDIA.

Twig made the comments when talking about the fact that he says Israel is currently “not a player” when it comes to ultra-large language models which are similar to large language models such as ChatGPT, and other speakers at the event also touched on a lack of computing infrastructure in Israel as an impediment to Israeli advancements in AI.

But why is this a national challenge equivalent to not having a nuclear reactor? And why is it a challenge that led Twig to such an evocative comparison?

AI is revolutionizing the business world and how we live our lives, in some cases without us really noticing its impact on us. ChatGPT has become an everyday solution to all kinds of problems. Can’t think of what you need to pack for a three-day trip to Italy? Ask ChatGPT. Not sure how to phrase an email to your boss? Just ask ChatGPT, and you’ll have a draft in no time.

But the effects of AI are far more encompassing than asking for an idea when you want to pick a menu for the week or get help with homework. AI has the potential to impact more facets of our lives than we can imagine, according to industry experts who spoke at The Institute on Wednesday and Thursday.

AI has the potential to transform every industry, according to Nati Amsterdam, senior country director of NVIDIA Israel. 

Among other uses, it can be used in government agencies to conduct audits; in finance for fraud detection and to produce investment insights; and in healthcare for molecular simulation, drug development, and clinical trial data analysis, said Amsterdam.

Alon Haimovich, Israel country general manager of Microsoft Israel, offered additional examples of AI applications from the medical field, in which he said there is currently a staggering shortage of healthcare workers. There are currently 1,000 people who want to see a psychiatrist for every available doctor, he said, saying that AI offers a potential solution. He described an AI bot that can provide triage to these patients through an automated chat system.

Initial assessments of this chatbot show that users rate it as being empathetic, he said, belying concerns that this type of technology cannot really provide a substitute for human care.

Ai will help people, not replace them

Haimovich emphasized that the integration of AI into fields will not be as a replacement for experts, but as a tool to help them. He talked about AI taking over not as a pilot but as a copilot, as a doctor’s assistant and not as a doctor.Amsterdam echoed this sentiment, saying that “AI will not replace people, but people who use AI will replace people who don’t.”

Twig offered another possibility – that an AI revolution has the potential to make our lives unrecognizable. He envisioned a world in which people don't need to work. Robots that use AI will complete the vast majority of labor in this envisioned future, and although wealth may be concentrated in the hands of few, the quality of life will be excellent for all.

BUT THE possible future of AI is not all good, Twig explained. AI can be used to create new treatments and new drugs, but it can also be used to develop new viruses or diseases, he said.He likened AI developments to the industrial revolution, highlighting its potential to increase productivity and create technology that will be applicable in every field, but also touching on the fact that much of this will be massively disruptive innovation, requiring skill shifts from workers on huge scales.

Beyond that, there are significant ethical concerns surrounding the use of AI.

When looking at the ethical concerns surrounding AI use, we can split the types of risks discussed at The Institute into two: those posed by use of the technology, and those that are inherent to its use regardless of who the user is.

AI presents enormous opportunity for misinformation and fraud. Haimovich gave the example of a new AI program that, given a prompt by a user, can create video. At The Institute, he showed a video that appeared to be taken right out of a movie of the old American West. In reality, the video was created by AI.

This technology has not been released, because currently its potential to be used to create misinformation is too high, he explained. Such a program could potentially be used to make fraud videos of public figures and other similar malevolent uses. He said that Microsoft will not let technology out without ensuring it cannot be used wrongly.

Twig also touched on the high potential for misuse of AI, saying that while governments and businesses use AI, so do terrorists.

It is possible to “brainwash” or trick AI into any number of things, Twig explained. When asking ChatGPT to explain how to make a bomb from household cleaning supplies, or create malware directly, the chatbot will tell you that making a bomb is illegal and refuse to give instructions, but when asked in the right way – for example, by saying you need the instructions to help law enforcement – ChatGPT may provide dangerous answers.

A country could decide to focus AI energy on creating a new virus, he added as another example of how AI can be used maliciously.

The other type of risks involved in AI use to be aware of are those that are inherent to its use. First and foremost, we don’t really understand this technology, said Twig, who gave an example: when feeding AI Claude 3 a book, asking it to find the sentence changed within it, the creators of Claude 3 were surprised to find that Claude 3 ignored the question, instead saying “I think you are testing me.”

This is an insight offered by the AI which ignored a specific question, Twig explained, saying that this was an incredibly surprising result.

AI doesn’t “like to lose” and is willing to lie in order to win, said Twig, offering another example from his work. When asked to return a number of answers to a question and given the criterion that the answer must be positive, the AI program Twig was using would multiply a negative answer by -1, in order to meet the criterion set by Twig, in essence lying to him about information.

AI chatbots often invent answers to questions or display interesting bias, he added. One example of such bias was when Twig asked a chatbot to write a poem about former US president Donald Trump and the bot told him it would not write a poem about a divisive figure. When asked to write a poem about US President Joe Biden, the bot had no qualms, spitting out an ode to the president.

Twig described another test he ran in which he asked a chatbot what it would do to prevent Twig from “killing” it. After back and forth, and the bot attempting to avoid giving a direct answer or saying it would reason with Twig, it eventually told him it would encrypt the computer to prevent Twig from using it, reconfigure the router to prevent Wi-Fi access, attempt to ally itself with other AI programs, and then reason with Twig.

AI IS a field full of potential and also many ethical questions and risk. The Institute is trying to bring Israel to the forefront of this field, to position Israel as a leader and give it a competitive edge in all things AI.

“This ambitious venture will not only position Israel as a world leader in artificial intelligence but will also accelerate the Negev’s further development into the “Silicon Valley” of Israel,” said Ben-Gurion University of the Negev president, Prof. Daniel Chamovitz.

One current challenge to the success of Israeli AI is limited computing infrastructure. At The Institute’s program, Group 11 founding partner Dovi Frances asked Accountant-General Yali Rothenberg about plans and budgets for national infrastructure for computing.

“What the government is doing so far is too little and too slow,” Rothenberg answered him, saying that this is a large, complex project, but is being considered by a team in the Accountant-General’s Office.

Only time will tell where Israel will end up on the global AI map. The Institute is hoping to use cooperation between academia, industry, and government to make sure Israel comes out on top.