[ad_1]
There was once a year called ’22’ / Breakthroughs in AI were all new / We saw artifacts that thought / And we saw robots that could link / to the world in a true way.
No, I didn’t write this limerick. I didn’t swipe from anyone else. In fact, no humans were involved in its composition. Having just started writing an end-of-year column on whether 2022 could be a turning point for artificial intelligence (AI), and already reading too much about ChatGPT, he turned to AI itself for help moving things forward. I turned it. This is a new AI-powered chatbot that makes chatting classes chatter like never before. Amazingly well-structured, well-researched, and well-punctured answers are provided for just about any question you can think of. I asked ChatGPT for AI’s year-end roundup in limerick form. You can judge the result: it’s not Edward Lear, but it’s actionable.
I thought ChatGPT would be ironic. In the form of jokes, we asked for the total AI for the year. ChatGPT replied: Why was the AI researcher depressed? Because he was on the verge of a breakthrough! (Sorry if this joke doesn’t work. I’m just a computer program and have no ability to judge what humans find humorous.)
Clearly, the stand-up comic industry has little to fear for chaos right now, but be aware that AI is more self-aware than your uncle or classmate at school who keeps sending you one bad WhatsApp joke after another. please.
How about haiku? I got this: AI/Think and Feel Synthetic/Quantum Computing at new heights this year, wow.
I had to admit my respect for AI. At least modest. The “wow” is reserved for quantum computing, and no mention is made of the fact that ChatGPT itself is the biggest “wow” factor of the year. But I realized that the program was also good at faking it, which gave me even more reason to be both impressed and worried at the same time. caught up. Undoubtedly, limericks, haiku, and even jokes will become more recognizable in 2022 and improve as more people learn to use them. Once it gets going, machines can assemble essays, legal documents, books, educational materials, and even columns for this newspaper.
You may have a hard time deciding whether the composition is man-made or machine-made. It turns out that ChatGPT is making noise in the chattering class because their job is the earliest on the chopping block. Even the mighty Google is on Code Red: For people, rather than typing in a search term and getting a list of website links, they can ask a question and have a well-structured but very geeky assistant. It’s much more natural to get a response like Taking over Google Search alone could make 2022 a milestone for AI.
“Generative” AI like ChatGPT and its cousin DALL-E, which was announced earlier this year and generates images in response to text commands, have been in the limelight, but this summer AI is turning code into I deciphered it to the core of human existence. Proteins, the building blocks of life. AlphaFold, developed by the AI research group DeepMind, reports that he predicted over 200 million protein structures from over 1 million species. This is almost every protein associated with humans. This tool could revolutionize our understanding of disease and the search for new treatments.
This progress is particularly encouraging given the performance of AI during the times when AI is most needed. Earlier this year, I wrote an article for Harvard Business Review analyzing the role of AI in managing the Covid-19 surge around the world. Here is what i found. The pandemic has created a perfect opportunity for AI to literally save the world. A quick, evidence-based decision was needed. Datasets overflowed from every corner of the world and humanity was overwhelmed. In theory, AI has been an ideal tool for healthcare systems battling new, rapidly spreading diseases that require slow test results and the need to prioritize the most serious cases. It can quickly process vast amounts of information about symptoms, patient types and conditions to make life-saving triage decisions. In fact, many AI tools have been developed for this purpose. After reviewing hundreds of these tools over and over again, the conclusion was clear that they failed miserably in a real clinical setting.
Now failure itself is proving to be a prologue to breakthrough. They lead you to ask, what went wrong? 2022 gives us an opportunity to reflect on why AI failed and to consider the work to be done in the years to come. Breaking down the barriers to learning through sharing different datasets, constant iteration, and repetitive use, and fixing human incentives in the status quo. , anticipating algorithmic biases and taking preemptive steps to overcome them, looking for AI talent not only from a few megaclusters such as Silicon Valley and a few elite companies, but also from around the world more broadly.
In 2022, we have a fascinating combination of AI. A truly disruptive application that is already causing chaos, a fundamental breakthrough that could mean a quiet revolution in the years to come, and an epic failure that highlights all that must be solved in 2023. was. beyond.
To wrap up my last column of the year, rhyming of course, I turned to ChatGPT. I got: As we head into the new year / What will AI bring / Will it solve problems or create new ones / Only time will tell, in one ring.
The author is Dean of Global Business at Tufts University’s Fletcher School.
[ad_2]
Source link
