• D.C.
  • BXL
  • Lagos
  • Riyadh
  • Beijing
  • SG
  • D.C.
  • BXL
  • Lagos
Semafor Logo
  • Riyadh
  • Beijing
  • SG


In this edition, we talk to filmmaker Nenad Cicin-Sain about his flopped AI screenwriter, a $1 milli͏‌  ͏‌  ͏‌  ͏‌  ͏‌  ͏‌ 
 
rotating globe
December 18, 2024
semafor

Technology

technology
Sign up for our free newsletters
 
Reed Albergotti
Reed Albergotti

Hi, and welcome back to Semafor Tech.

ChatGPT and AI played a big part in the strikes over the last year by Hollywood writers and actors. But when one screenwriter tried putting ChatGPT to the test, he got hilarious results that say a lot about where the technology is today and where it’s headed. Spoiler: It’s disruptive, but not in the ways people assume.

It may be that AI doesn’t replace artists, but instead changes who is able to create art. And I’m starting to think that may be true across the board when it comes to AI.

The internet did something similar. Television personalities suddenly didn’t need to pay their dues working their way up the ranks of big networks. They just needed an internet connection and a camera. A global opinion leader went from being a salaried job at a newspaper to anyone with a laptop who had something to say.

What happens when even more barriers come down? What if creating great software, which once required many hours of obsessive practice, is accessible to anyone with a good product idea?

There is a possibility that AI unlocks human creativity, rather than replacing it. That’s actually a very disruptive idea that should probably worry people with power more than it scares artists.

And below, a $1 million prize for anyone who can train an AI model to score 90% on an impossible-to-cheat coding test.

Move Fast/Break Things
A chart showing the capital raised by AI startups, with the figure steadily increasing over 10 years

➚ MOVE FAST: Bricks. In another sign of the AI boom, startup Databricks has raised $10 billion at a $62 billion valuation. That’s a massive amount at a time when $6 billion is becoming the norm for the likes of OpenAI and xAI.

➘ BREAK THINGS: Clicks. US authorities may ban the most popular home-internet routers made by TP-Link, which are tied to hacks allegedly perpetrated by China, The Wall Street Journal reported. But with time running out in the Biden administration, that decision may fall to Donald Trump’s officials.

PostEmail
Artificial Flavor
An Oreo cookie
rawpixel.com

Integrating AI into business is sweet.

Oreo-maker Mondelez has started applying AI to its research and development processes to more quickly and efficiently create the snacks consumers want, The Wall Street Journal reported. The machine learning tool draws on Mondelez’s recipes for fan favorites like Sour Patch Kids and Chips Ahoy!, suggesting both new products and updates to existing recipes — alleviating the previous trial-and-error workflow. The carb creations make it into production trials four to five times more quickly, Mondelez told the Journal.

Companies are racing to iterate their products and processes with AI in out-of-the-box ways and broadcast their accomplishments to please investors. Mondelez’s stock is down over the last year as it continues to face pressure over evolving consumer tastes and fierce competition. The AI tool, developed in 2019, has been used in more than 70 projects, the company said. So next time you bite into an Oreo, don’t be surprised if you taste a hint of silicon.

PostEmail
Reed Albergotti

Why filmmaker Nenad Cicin-Sain fired his AI screenwriter

THE SCENE

Filmmaker Nenad Cicin-Sain’s latest project, a drama about a politician who relies entirely on artificial intelligence to make decisions for him, stalled recently because a screenwriter kept coming up with lame excuses for missing deadlines. The name of the deadbeat scribe: ChatGPT.

Like many in the movie business, Cicin-Sain had been thinking a lot about how AI might disrupt his industry, which was recently roiled by writer and actor strikes over such concerns. Earlier this year, a film executive told him he couldn’t wait for the next version of ChatGPT, which he expected to be good enough to replace screenwriters entirely.

“I wanted to become as knowledgeable as possible,” said Cicin-Sain, whose recent documentary about the underground music scene in war-torn Sarajevo was produced by actors Ben Affleck and Matt Damon.

So he set out to use AI to write his next screenplay, with the goal of turning the experience into a documentary or a feature film. “I expected it to instantaneously pump out a screenplay once I created all the prompts,” he said.

Instead, what ensued was a bizarre episode lasting weeks in which ChatGPT kept asking for more time, continuously changing deadlines and then coming up with reasons why it couldn’t meet them.

The episode appears to be a new kind of elaborate, multi-day “hallucination,” where an AI model answers something incorrectly and then refuses to budge from that position.

Director Nenad Cicin-Sain at the press conference for the Berlinale film “Kiss the Future” in 2023.
Soeren Stache/DPA/Picture Alliance

The first clue was when the chatbot asked for two to four weeks to complete the project. Cicin-Sain, thinking perhaps that’s how long it takes to process a complex task like a screenplay, asked if it could be done in two weeks. ChatGPT agreed.

“I’ll make sure to update you at the end of each day with the progress on the screenplay’s outline and scene breakdown. Looking forward to working on this with you!” it wrote.

But it missed the deadline. “You never updated me,” Cicin-Sain wrote the next day.

ChatGPT was apologetic, promising to get the job done. But the next day, no update came. Cicin-Sain kept following up, and each time there was a new excuse.

Then came the gaslighting. ChatGPT claimed the deadline had not been clear.

“Looking back at our conversations, I believe this is the first instance where I gave a specific timeline for delivering a draft. Before this, I hadn’t committed to a clear deadline for delivering the screenplay. Most of our discussions were focused on brainstorming, outlining, and developing characters and themes,” it wrote.

“But can you look back now at all our conversations and give me the exact number of times I asked you for a delivery timeline and the exact number of times you didn’t meet it,” Cicin-Sain wrote.

At this point, Cicin-Sain had given up on the idea that ChatGPT was going to write him a screenplay. So he started a new chat from scratch, offering up a different screenplay idea. He got the same result.

In the end, Cicin-Sain came to the conclusion that there’s no reason to fear that AI will take over screenwriting for at least a handful of years.

Read what ChatGPT did when Cicin-Sain asked it to write a scene from There Will be Blood. →

PostEmail
Obsessions
A person working on code at a laptop
rawpixel.com

It pays to play with AI models. In this case, developers can get a cash prize for creating an open-source AI model that performs well on a new benchmark to test software coding. Andy Konwinski, a co-founder of Databricks and Perplexity, is offering up to $1 million for anyone who can get a model to score up to 90% on the new benchmark.

Konwinski’s contest, called K Prize, is the latest in a wave of similar awards for people who can solve problems with AI.

The long-standing problem with AI benchmarks is that the training data used by AI models is often contaminated with the benchmarks themselves. It’s like giving someone the answers to a test before they take it. Despite the grade inflation, a coding assessment called SWE-bench said on its website that the highest-performing model currently scores 55% on the evaluation, which tests real-world software problems posted to the popular coding repository site GitHub.

The top submission to the K Prize will receive at least $50,000, even if it doesn’t hit the 90% mark. “The goal isn’t necessarily to win the world championship of this thing,” Konwinski told Semafor. “It’s just to catalyze energy and breakthroughs.”

Read the full interview with Konwinski and more on why he set up the prize. →

PostEmail
What We’re Tracking

TikTok CEO Shou Zi Chew is pulling every lever he has to overturn the US ban on his social media platform. This week, that means engaging the highest court in the US and appealing straight to the president-elect. The Supreme Court said Wednesday it will hear the case that the ban violates the First Amendment, giving TikTok another fighting chance after an Appeals court upheld the law earlier this month. But TikTok isn’t out of the water — it will have to present its case to SCOTUS on Jan. 10, just nine days before the ban takes effect.

Aside from the SCOTUS decision, TikTok may have an unlikely ally in Trump, who sought a ban on the app just four years ago. However, Trump’s TikTok stardom during election season may have bought the app some kudos points with the future president, who credits some of his win to young voters on the platform. “I have a warm spot in my heart for TikTok,” he said Monday, the same day he met with Chew at Mar-a-Lago, The Hill reported. He will “take a look” at preventing the ban from taking place, he added.

A chart showing the percentage of adults who support or oppose a TikTok ban, with the percentage of those supporting it dropping sharply

This time around, it could benefit Trump to differentiate himself and stand by his campaign promise to “save” the app. Keeping TikTok around also puts pressure on Meta and its CEO Mark Zuckerberg, who Trump has taken shots at over the platform’s alleged intervention in the 2020 election. Meta has attempted to make peace with a $1 million donation to Trump’s inaugural fund (as did Amazon and OpenAI’s Sam Altman). Don’t be surprised if you also see a hefty contribution from TikTok, as that appears to be the ticket to play.

Even with Trump’s support, TikTok still has hurdles to cross. Trump alone can’t make the legal ban go poof. If TikTok does sell to a US entity, which its parent company previously said it had no plans to do, the app would continue to operate in the US but potentially without the secret sauce (its algorithm) that made TikTok so popular. Billionaire Frank McCourt is mounting a bid for the app sans algorithm. Shark Tank’s Kevin O’Leary, former Treasury Secretary Steven Mnuchin, and former Activision Blizzard chief Bobby Kotick have also expressed interest.

PostEmail
Semafor Spotlight
A graphic saying “A great read from Semafor Net Zero”A liquefied natural gas (LNG) tanker is tugged towards a thermal power station
Issei Kato/Reuters

A top US energy expert thinks the Biden administration missed the mark in its long-awaited study of US natural gas exports, Semafor’s Tim McDonnell reported, arguing that pushback on a key element of President-elect Donald Trump’s fossil fuel ambitions will have negative consequences.

Expanding LNG exports, on the other hand, is an unalloyed boon to both the US economy and national security, analysts argued.

For more on the energy transition, subscribe to Semafor’s Net Zero newsletter. →

PostEmail